Move agent section to fab (#6597)

Co-authored-by: Anthony Stirling <77850077+Frooodle@users.noreply.github.com>

.github/workflows/docker-compose-tests-saas.yml

@@ -0,0 +1,136 @@
+name: Docker Compose Cucumber tests (saas / PAYG)
+
+# Self-contained CI job for the PAYG shadow-mode cucumber scenarios.
+# Triggers only on PAYG-relevant paths so we don't add CI minutes to every PR
+# that doesn't touch the saas flavour.
+#
+# Companion to `docker-compose-tests.yml` (which runs against the
+# proprietary-flavour stack and skips features/payg via behave.ini's
+# exclude_re). Kept as a separate workflow so the saas matrix can fail and
+# succeed independently without touching the main cucumber harness.
+
+on:
+  pull_request:
+    paths:
+      - "app/saas/**"
+      - "testing/cucumber/features/payg/**"
+      - "testing/cucumber/features/steps/payg_step_definitions.py"
+      - "testing/cucumber/requirements.txt"
+      - "testing/compose/docker-compose-saas.yml"
+      - "testing/compose/payg/**"
+      - "testing/test-payg.sh"
+      - ".github/workflows/docker-compose-tests-saas.yml"
+  push:
+    branches: [main]
+    paths:
+      - "app/saas/**"
+      - "testing/cucumber/features/payg/**"
+      - "testing/cucumber/features/steps/payg_step_definitions.py"
+      - "testing/cucumber/requirements.txt"
+      - "testing/compose/docker-compose-saas.yml"
+      - "testing/compose/payg/**"
+      - "testing/test-payg.sh"
+      - ".github/workflows/docker-compose-tests-saas.yml"
+
+permissions:
+  contents: read
+
+jobs:
+  pick:
+    uses: ./.github/workflows/_runner-pick.yml
+
+  docker-compose-tests-saas:
+    needs: pick
+    runs-on: ${{ needs.pick.outputs.is_fork == 'true' && 'ubuntu-latest' || 'depot-ubuntu-24.04-4' }}
+    permissions:
+      actions: write
+      contents: read
+      checks: write
+    env:
+      DEPOT_TOKEN: ${{ secrets.DEPOT_TOKEN }}
+
+    steps:
+      - name: Harden Runner
+        uses: step-security/harden-runner@ab7a9404c0f3da075243ca237b5fac12c98deaa5 # v2.19.3
+        with:
+          egress-policy: audit
+
+      - name: Checkout Repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+      - name: Set up JDK 25
+        uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
+        with:
+          java-version: "25"
+          distribution: "temurin"
+
+      - name: Cache Gradle dependency artifacts
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
+        with:
+          path: |
+            ~/.gradle/wrapper
+            ~/.gradle/caches/modules-2/files-2.1
+            ~/.gradle/caches/modules-2/metadata-2.*
+          key: gradle-deps-saas-${{ runner.os }}-jdk-25-${{ hashFiles('**/gradle/wrapper/gradle-wrapper.properties', '**/*.gradle', '**/*.gradle.kts', 'settings.gradle', 'settings.gradle.kts', 'gradle/libs.versions.toml') }}
+
+      - name: Setup Gradle
+        uses: gradle/actions/setup-gradle@50e97c2cd7a37755bbfafc9c5b7cafaece252f6e # v6.1.0
+        with:
+          gradle-version: 9.3.1
+          cache-disabled: true
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd # v4.0.0
+
+      - name: Expose GitHub runtime for Buildx cache
+        uses: crazy-max/ghaction-github-runtime@04d248b84655b509d8c44dc1d6f990c879747487 # v4.0.0
+
+      # No "Install Docker Compose" step: Ubuntu runners ship with `docker compose`
+      # v2 (built into the Docker CLI). test-payg.sh uses the v2 form throughout
+      # (`docker compose …`, no hyphen), so the legacy v1 `docker-compose` binary
+      # isn't needed. Avoids a `curl | sudo install` without checksum verification
+      # (Aikido flagged this when copy-pasted from docker-compose-tests.yml).
+
+      - name: Set up Python
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.12"
+          cache: "pip"
+          cache-dependency-path: ./testing/cucumber/requirements.txt
+
+      - name: Pip requirements
+        run: |
+          pip install --require-hashes --only-binary=:all: -r ./testing/cucumber/requirements.txt
+
+      - name: Run PAYG Cucumber Tests
+        env:
+          MAVEN_USER: ${{ secrets.MAVEN_USER }}
+          MAVEN_PASSWORD: ${{ secrets.MAVEN_PASSWORD }}
+          MAVEN_PUBLIC_URL: ${{ secrets.MAVEN_PUBLIC_URL }}
+        run: |
+          chmod +x ./testing/test-payg.sh
+          ./testing/test-payg.sh
+
+      - name: Dump saas container logs on failure
+        if: failure()
+        run: |
+          docker compose -f testing/compose/docker-compose-saas.yml logs --tail 500 stirling-pdf-saas || true
+          docker compose -f testing/compose/docker-compose-saas.yml logs --tail 200 postgres-saas || true
+
+      - name: Upload PAYG Cucumber Report
+        if: always()
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: payg-cucumber-report
+          path: testing/cucumber/report-payg.html
+          retention-days: 7
+          if-no-files-found: warn
+
+      - name: PAYG Cucumber Test Report
+        if: always()
+        uses: dorny/test-reporter@a43b3a5f7366b97d083190328d2c652e1a8b6aa2 # v3.0.0
+        with:
+          name: PAYG Cucumber Tests
+          path: testing/cucumber/junit-payg/*.xml
+          reporter: java-junit
+          fail-on-error: false

.gitleaksignore

@@ -3,3 +3,12 @@
 # intentionally in #6150 so engine/.env has a working default, with real
 # credentials overridden via engine/.env.local.
 engine/.env:generic-api-key:41
+
+# MCP test fixtures / harness - no real secrets:
+# - test-only API key constant in an integration test
+# - JDBC URL + throwaway Keycloak creds in the local test compose
+# - placeholder / shell-variable Bearer headers in curl-based validation scripts
+app/proprietary/src/test/java/stirling/software/proprietary/mcp/security/McpApiKeyIntegrationTest.java:generic-api-key:40
+testing/compose/docker-compose-keycloak-mcp.yml:generic-api-key:25
+testing/compose/validate-mcp-apikey.sh:curl-auth-header:73
+testing/compose/validate-mcp-test.sh:curl-auth-header:92

.taskfiles/desktop.yml

@@ -1,7 +1,9 @@
 version: '3'
 
 vars:
-  JLINK_MODULES: "java.base,java.compiler,java.desktop,java.instrument,java.logging,java.management,java.naming,java.net.http,java.prefs,java.rmi,java.scripting,java.security.jgss,java.security.sasl,java.sql,java.transaction.xa,java.xml,java.xml.crypto,jdk.crypto.ec,jdk.crypto.cryptoki,jdk.unsupported"
+  # jdk.dynalink is required by VeraPDF (PDF/A validation); without it the bundled JRE throws
+  # NoClassDefFoundError: jdk/dynalink/Namespace at runtime in get-info-on-pdf and verify-pdf
+  JLINK_MODULES: "java.base,java.compiler,java.desktop,java.instrument,java.logging,java.management,java.naming,java.net.http,java.prefs,java.rmi,java.scripting,java.security.jgss,java.security.sasl,java.sql,java.transaction.xa,java.xml,java.xml.crypto,jdk.crypto.ec,jdk.crypto.cryptoki,jdk.unsupported,jdk.dynalink"
 
   # Override via JPDFIUM_PLATFORMS env (csv of platform keys, or 'all').
   JPDFIUM_PLATFORMS:
@@ -62,21 +64,21 @@ tasks:
     deps: [prepare]
     dir: editor
     cmds:
-      - npx tauri build --bundles app
+      - npx tauri build --bundles app --config '{"bundle":{"createUpdaterArtifacts":false}}'
 
   build:dev:windows:
     desc: "Build Tauri desktop NSIS installer (Windows)"
     deps: [prepare]
     dir: editor
     cmds:
-      - npx tauri build --bundles nsis
+      - npx tauri build --bundles nsis --config '{"bundle":{"createUpdaterArtifacts":false}}'
 
   build:dev:linux:
     desc: "Build Tauri desktop AppImage (Linux)"
     deps: [prepare]
     dir: editor
     cmds:
-      - npx tauri build --bundles appimage
+      - npx tauri build --bundles appimage --config '{"bundle":{"createUpdaterArtifacts":false}}'
 
   test:
     desc: "Run Tauri/Cargo tests"

.taskfiles/e2e.yml

@@ -212,3 +212,55 @@ tasks:
     desc: "Stop the SAML keycloak test environment"
     cmds:
       - docker compose -f testing/compose/docker-compose-keycloak-saml.yml down -v
+
+  mcp:up:
+    desc: "Start the MCP keycloak test environment (Stirling as OAuth resource server)"
+    summary: |
+      Brings up Keycloak (OAuth authorization server) + Stirling configured as an
+      MCP resource server, then you can exercise /mcp with real Keycloak tokens.
+      Set LICENSE_KEY=<KEY> to skip the interactive license prompt:
+        task e2e:mcp:up LICENSE_KEY=abc123
+      Pass extra flags via -- :
+        task e2e:mcp:up -- --validate --nobuild
+    ignore_error: true
+    cmds:
+      - bash testing/compose/start-mcp-test.sh {{if .LICENSE_KEY}}--license-key "{{.LICENSE_KEY}}"{{end}} {{.CLI_ARGS}}
+
+  mcp:manual:
+    desc: "Start the MCP keycloak test env in manual mode (prints URLs + a live token for your client)"
+    summary: |
+      Brings the stack up and prints copy-paste URLs/commands plus a freshly minted
+      access token so you can drive your own MCP client (Inspector, curl, ...).
+        task e2e:mcp:manual LICENSE_KEY=<your-license-key>
+      Add --nobuild if the images are already built:
+        task e2e:mcp:manual LICENSE_KEY=<your-license-key> -- --nobuild
+    ignore_error: true
+    cmds:
+      - bash testing/compose/start-mcp-test.sh --manual {{if .LICENSE_KEY}}--license-key "{{.LICENSE_KEY}}"{{end}} {{.CLI_ARGS}}
+
+  mcp:apikey:
+    desc: "Start the MCP test env in API-KEY manual mode (no OAuth/IdP): mints a key + prints client settings"
+    summary: |
+      Brings Stirling up in apikey auth mode and prints copy-paste client settings with a freshly
+      minted X-API-KEY - ideal for clients whose OAuth layer can't reach localhost.
+        task e2e:mcp:apikey LICENSE_KEY=<your-license-key>
+      Add --nobuild if images are already built:
+        task e2e:mcp:apikey LICENSE_KEY=<your-license-key> -- --nobuild
+    ignore_error: true
+    cmds:
+      - bash testing/compose/start-mcp-test.sh --apikey {{if .LICENSE_KEY}}--license-key "{{.LICENSE_KEY}}"{{end}} {{.CLI_ARGS}}
+
+  mcp:validate:
+    desc: "Validate the running MCP keycloak test environment end-to-end (oauth mode + real MCP SDK client)"
+    cmds:
+      - bash testing/compose/validate-mcp-test.sh
+
+  mcp:validate-apikey:
+    desc: "Validate the MCP server in API-KEY auth mode (mints a key + real MCP SDK client), then restore oauth"
+    cmds:
+      - bash testing/compose/validate-mcp-apikey.sh
+
+  mcp:down:
+    desc: "Stop the MCP keycloak test environment"
+    cmds:
+      - docker compose -f testing/compose/docker-compose-keycloak-mcp.yml down -v

.taskfiles/engine.yml

@@ -1,5 +1,12 @@
 version: '3'
 
+vars:
+  # Engine-specific names to avoid overriding the root Taskfile's FIND_FREE_PORT_*
+  # vars (Task merges included-file vars into the global scope).
+  # Paths are relative to the engine/ include dir.
+  ENGINE_FIND_FREE_PORT_SH: "bash ../scripts/find-free-port.sh"
+  ENGINE_FIND_FREE_PORT_PS: "powershell -NoProfile -File ../scripts/find-free-port.ps1"
+
 tasks:
   install:
     desc: "Install engine dependencies"
@@ -29,7 +36,10 @@ tasks:
     ignore_error: true
     dir: src
     vars:
-      PORT: '{{.PORT | default "5001"}}'
+      # When PORT is provided (e.g. from dev:all), use it directly.
+      # When running standalone, probe for a free port starting at 5001.
+      PORT:
+        sh: '{{if .PORT}}echo {{.PORT}}{{else if eq OS "windows"}}{{.ENGINE_FIND_FREE_PORT_PS}} 5001{{else}}{{.ENGINE_FIND_FREE_PORT_SH}} 5001{{end}}'
     env:
       PYTHONUNBUFFERED: "1"
     cmds:
@@ -41,7 +51,10 @@ tasks:
     ignore_error: true
     dir: src
     vars:
-      PORT: '{{.PORT | default "5001"}}'
+      # When PORT is provided (e.g. from dev:all), use it directly.
+      # When running standalone, probe for a free port starting at 5001.
+      PORT:
+        sh: '{{if .PORT}}echo {{.PORT}}{{else if eq OS "windows"}}{{.ENGINE_FIND_FREE_PORT_PS}} 5001{{else}}{{.ENGINE_FIND_FREE_PORT_SH}} 5001{{end}}'
     env:
       PYTHONUNBUFFERED: "1"
     cmds:

.taskfiles/frontend.yml

@@ -327,19 +327,19 @@ tasks:
 
   test:
     desc: "Run tests"
-    deps: [install]
+    deps: [prepare]
     cmds:
       - npx vitest run --root editor
 
   test:watch:
     desc: "Run tests in watch mode"
-    deps: [install]
+    deps: [prepare]
     cmds:
       - npx vitest --watch --root editor
 
   test:coverage:
     desc: "Run tests with coverage (one-shot; CI-friendly)."
-    deps: [install]
+    deps: [prepare]
     cmds:
       # `vitest run` makes this CI-safe (the bare `vitest` form enters watch
       # mode). Explicit reporter list because v8 + json-summary is what the

app/common/src/main/java/stirling/software/SPDF/pdf/parser/CompositeTableParser.java

@@ -1,73 +0,0 @@
-package stirling.software.SPDF.pdf.parser;
-
-import static stirling.software.SPDF.pdf.parser.PdfModels.*;
-
-import java.io.IOException;
-import java.util.List;
-
-import org.apache.pdfbox.pdmodel.PDDocument;
-import org.springframework.context.annotation.Primary;
-import org.springframework.stereotype.Service;
-
-import lombok.RequiredArgsConstructor;
-import lombok.extern.slf4j.Slf4j;
-
-/**
- * Chains table parsers in priority order: Tabula lattice → Tabula stream → {@link
- * LineAlignmentTableParser}. The first parser returning a result above {@link
- * #TABULA_CONFIDENCE_THRESHOLD} wins; results from different parsers are never mixed on one page.
- */
-@Service
-@Primary
-@RequiredArgsConstructor
-@Slf4j
-public class CompositeTableParser implements TableParser {
-
-    /** Min Tabula confidence to accept results; below this LineAlignment is tried instead. */
-    static final float TABULA_CONFIDENCE_THRESHOLD = 0.5f;
-
-    private final TabulaTableParser tabulaParser;
-    private final LineAlignmentTableParser lineAlignmentParser;
-
-    @Override
-    public List<TableFragment> parse(PDDocument document, RawPage rawPage) throws IOException {
-        // Step 1: Tabula lattice mode (ruled/bordered tables).
-        List<TableFragment> latticeResults = filterConfident(tabulaParser.parse(document, rawPage));
-        if (!latticeResults.isEmpty()) {
-            log.debug(
-                    "Page {}: using Tabula lattice ({} table(s))",
-                    rawPage.pageNumber(),
-                    latticeResults.size());
-            return latticeResults;
-        }
-
-        // Step 2: Tabula stream mode (borderless/whitespace-delimited tables).
-        // parseStream is not on the TableParser interface — this intentionally couples to the
-        // concrete TabulaTableParser since stream mode is a Tabula-specific concept.
-        List<TableFragment> streamResults =
-                filterConfident(tabulaParser.parseStream(document, rawPage));
-        if (!streamResults.isEmpty()) {
-            log.debug(
-                    "Page {}: using Tabula stream ({} table(s))",
-                    rawPage.pageNumber(),
-                    streamResults.size());
-            return streamResults;
-        }
-
-        // Step 3: Geometry-based line-alignment fallback.
-        List<TableFragment> lineResults = lineAlignmentParser.parse(document, rawPage);
-        if (!lineResults.isEmpty()) {
-            log.debug(
-                    "Page {}: using LineAlignment ({} table(s))",
-                    rawPage.pageNumber(),
-                    lineResults.size());
-            return lineResults;
-        }
-
-        return List.of();
-    }
-
-    private List<TableFragment> filterConfident(List<TableFragment> tables) {
-        return tables.stream().filter(t -> t.confidence() >= TABULA_CONFIDENCE_THRESHOLD).toList();
-    }
-}

app/common/src/main/java/stirling/software/SPDF/pdf/parser/LineAlignmentTableParser.java

@@ -1,528 +0,0 @@
-package stirling.software.SPDF.pdf.parser;
-
-import static stirling.software.SPDF.pdf.parser.PdfModels.*;
-
-import java.io.IOException;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.Collections;
-import java.util.Comparator;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
-import java.util.Optional;
-import java.util.TreeMap;
-import java.util.regex.Pattern;
-
-import org.apache.pdfbox.pdmodel.PDDocument;
-import org.springframework.stereotype.Service;
-
-import lombok.extern.slf4j.Slf4j;
-
-/**
- * Fallback {@link TableParser} for borderless financial tables using text geometry.
- *
- * <p>Identifies "anchor lines" (≥2 numeric tokens), builds a column grid from their right-edge
- * positions, groups vertically proximate anchor lines into table candidates, then scores each group
- * on column consistency and anchor density (confidence ceiling 0.85).
- */
-@Service
-@Slf4j
-public class LineAlignmentTableParser implements TableParser {
-
-    /** Width in points of each column position bucket. */
-    static final float COLUMN_BUCKET_PT = 5f;
-
-    /** Tolerance in buckets when matching a token's right-edge to a confirmed column position. */
-    private static final int COLUMN_MATCH_BUCKETS = 2;
-
-    /** Maximum gap (as a multiple of modal line spacing) before splitting a group. */
-    private static final float MAX_GAP_FACTOR = 2.5f;
-
-    /** Minimum anchor rows (numeric-heavy) to form a valid table. */
-    static final int MIN_TABLE_ROWS = 3;
-
-    /** Minimum confirmed column positions to form a valid table. */
-    static final int MIN_COLUMNS = 2;
-
-    /**
-     * Min fraction of anchor lines a column must appear on to be confirmed (permissive for N/A
-     * rows).
-     */
-    private static final double COLUMN_MIN_FREQUENCY = 0.40;
-
-    /**
-     * Matches financial numeric tokens: integers, decimals, parenthetical negatives, currency,
-     * percent, nil dashes.
-     */
-    private static final Pattern NUMERIC =
-            Pattern.compile("^[\\(\\-\\$£€¥]?\\d[\\d,\\.]*[\\)%]?$|^[-–—]$");
-
-    /**
-     * Lines within this y-distance are merged into one row (restores rows split by LineBuilder's
-     * column-gap logic).
-     */
-    static final float ROW_MERGE_TOLERANCE_PT = 2f;
-
-    // ── public API ───────────────────────────────────────────────────────────────────────────────
-
-    @Override
-    public List<TableFragment> parse(PDDocument document, RawPage rawPage) throws IOException {
-        List<RawLine> lines = rawPage.lines();
-        if (lines.size() < MIN_TABLE_ROWS) return List.of();
-
-        float modalSpacing = computeModalSpacing(lines);
-        List<TokenizedLine> tokenized =
-                mergeCoincidentLines(lines.stream().map(this::tokenize).toList());
-
-        List<TokenizedLine> anchors = tokenized.stream().filter(TokenizedLine::isAnchor).toList();
-
-        if (anchors.size() < MIN_TABLE_ROWS) return List.of();
-
-        List<Float> columnGrid = buildColumnGrid(anchors);
-        if (columnGrid.size() < MIN_COLUMNS) {
-            log.debug(
-                    "Page {}: LineAlignment — fewer than {} confirmed columns, skipping",
-                    rawPage.pageNumber(),
-                    MIN_COLUMNS);
-            return List.of();
-        }
-
-        List<List<TokenizedLine>> groups = groupRows(tokenized, columnGrid, modalSpacing);
-
-        List<TableFragment> results = new ArrayList<>();
-        for (int i = 0; i < groups.size(); i++) {
-            buildFragment(groups.get(i), columnGrid, rawPage.pageNumber(), i)
-                    .ifPresent(results::add);
-        }
-
-        log.debug(
-                "Page {}: LineAlignment detected {} table(s) ({} anchor lines, {} columns)",
-                rawPage.pageNumber(),
-                results.size(),
-                anchors.size(),
-                columnGrid.size());
-        return results;
-    }
-
-    // ── coincident-line merging ──────────────────────────────────────────────────────────────────
-
-    /**
-     * Merges tokenised lines sharing the same y-position into one row, rejoining label/value halves
-     * split by LineBuilder.
-     */
-    List<TokenizedLine> mergeCoincidentLines(List<TokenizedLine> tokenized) {
-        if (tokenized.size() < 2) return tokenized;
-
-        List<TokenizedLine> result = new ArrayList<>();
-        int i = 0;
-
-        while (i < tokenized.size()) {
-            float baseY = tokenized.get(i).line().bounds().y();
-            int j = i + 1;
-            while (j < tokenized.size()
-                    && Math.abs(tokenized.get(j).line().bounds().y() - baseY)
-                            <= ROW_MERGE_TOLERANCE_PT) {
-                j++;
-            }
-
-            if (j == i + 1) {
-                result.add(tokenized.get(i));
-            } else {
-                result.add(mergeGroup(tokenized.subList(i, j)));
-            }
-            i = j;
-        }
-
-        return result;
-    }
-
-    private TokenizedLine mergeGroup(List<TokenizedLine> group) {
-        List<TextFragment> mergedFragments =
-                group.stream()
-                        .flatMap(tl -> tl.line().fragments().stream())
-                        .sorted(Comparator.comparingDouble(f -> f.bounds().x()))
-                        .toList();
-
-        Bounds mergedBounds =
-                group.stream()
-                        .map(tl -> tl.line().bounds())
-                        .reduce(Bounds::merge)
-                        .orElse(group.get(0).line().bounds());
-
-        RawLine mergedLine =
-                new RawLine(
-                        group.get(0).line().lineId(),
-                        mergedFragments,
-                        mergedBounds,
-                        group.get(0).line().pageNumber());
-
-        return tokenize(mergedLine);
-    }
-
-    // ── tokenisation ─────────────────────────────────────────────────────────────────────────────
-
-    /**
-     * Splits fragments into word-level tokens; x-positions are estimated linearly within each
-     * fragment.
-     */
-    TokenizedLine tokenize(RawLine line) {
-        List<LineToken> tokens = new ArrayList<>();
-        for (TextFragment frag : line.fragments()) {
-            tokens.addAll(tokensFromFragment(frag));
-        }
-        List<LineToken> numeric = tokens.stream().filter(LineToken::numeric).toList();
-        return new TokenizedLine(line, tokens, numeric);
-    }
-
-    private List<LineToken> tokensFromFragment(TextFragment frag) {
-        String raw = frag.text();
-        if (raw == null || raw.isBlank()) return List.of();
-
-        float fragX = frag.bounds().x();
-        float fragWidth = frag.bounds().width();
-        int rawLen = raw.length();
-
-        List<LineToken> result = new ArrayList<>();
-        int offset = 0;
-        for (String part : raw.split("\\s+")) {
-            if (part.isEmpty()) {
-                offset++;
-                continue;
-            }
-            int idx = raw.indexOf(part, offset);
-            if (idx < 0) idx = offset;
-
-            float tokenX = rawLen > 0 ? fragX + ((float) idx / rawLen) * fragWidth : fragX;
-            float tokenRight =
-                    rawLen > 0
-                            ? fragX + ((float) (idx + part.length()) / rawLen) * fragWidth
-                            : fragX + fragWidth;
-
-            result.add(new LineToken(part, tokenX, tokenRight, NUMERIC.matcher(part).matches()));
-            offset = idx + part.length();
-        }
-        return result;
-    }
-
-    // ── column grid ──────────────────────────────────────────────────────────────────────────────
-
-    /**
-     * Returns confirmed column right-edge positions — those appearing on ≥ {@value
-     * #COLUMN_MIN_FREQUENCY} × N anchor lines.
-     */
-    private List<Float> buildColumnGrid(List<TokenizedLine> anchors) {
-        // bucket → set of line indices that contributed a numeric token to that bucket
-        Map<Integer, List<Integer>> bucketLines = new HashMap<>();
-        for (int i = 0; i < anchors.size(); i++) {
-            for (LineToken t : anchors.get(i).numeric()) {
-                int bucket = bucket(t.right());
-                bucketLines.computeIfAbsent(bucket, k -> new ArrayList<>()).add(i);
-            }
-        }
-
-        int minHits =
-                Math.max(MIN_TABLE_ROWS, (int) Math.ceil(anchors.size() * COLUMN_MIN_FREQUENCY));
-
-        // Confirmed buckets → average right-edge for that bucket
-        TreeMap<Integer, Float> confirmed = new TreeMap<>();
-        for (Map.Entry<Integer, List<Integer>> entry : bucketLines.entrySet()) {
-            // Count distinct lines
-            long distinctLines = entry.getValue().stream().distinct().count();
-            if (distinctLines >= minHits) {
-                double avg =
-                        entry.getValue().stream()
-                                .distinct() // weight each line equally regardless of token count
-                                .mapToDouble(
-                                        lineIdx ->
-                                                avgRightEdgeForBucket(
-                                                        anchors, lineIdx, entry.getKey()))
-                                .average()
-                                .orElse(entry.getKey() * (double) COLUMN_BUCKET_PT);
-                confirmed.put(entry.getKey(), (float) avg);
-            }
-        }
-
-        return new ArrayList<>(confirmed.values()); // already sorted by bucket (left to right)
-    }
-
-    /**
-     * Returns the average right-edge position of tokens in {@code line} whose bucket matches {@code
-     * targetBucket}, falling back to the bucket's nominal centre when no tokens match.
-     */
-    private double avgRightEdgeForBucket(
-            List<TokenizedLine> anchors, int lineIdx, int targetBucket) {
-        return anchors.get(lineIdx).numeric().stream()
-                .filter(t -> bucket(t.right()) == targetBucket)
-                .mapToDouble(LineToken::right)
-                .average()
-                .orElse(targetBucket * (double) COLUMN_BUCKET_PT);
-    }
-
-    // ── grouping ─────────────────────────────────────────────────────────────────────────────────
-
-    /**
-     * Groups anchor lines into table candidates, including adjacent label rows; a gap &gt;
-     * MAX_GAP_FACTOR × modal spacing splits groups.
-     */
-    private List<List<TokenizedLine>> groupRows(
-            List<TokenizedLine> all, List<Float> columnGrid, float modalSpacing) {
-        float maxGap = modalSpacing > 0 ? modalSpacing * MAX_GAP_FACTOR : 30f;
-
-        List<List<TokenizedLine>> groups = new ArrayList<>();
-        List<TokenizedLine> current = new ArrayList<>();
-
-        for (int i = 0; i < all.size(); i++) {
-            TokenizedLine tl = all.get(i);
-            boolean fits = tl.isAnchor() && matchesGrid(tl, columnGrid);
-
-            if (current.isEmpty()) {
-                if (fits) current.add(tl);
-                continue;
-            }
-
-            float gap =
-                    tl.line().bounds().y()
-                            - current.get(current.size() - 1).line().bounds().bottom();
-
-            if (gap > maxGap) {
-                groups.add(current);
-                current = new ArrayList<>();
-                if (fits) current.add(tl);
-                continue;
-            }
-
-            if (fits) {
-                current.add(tl);
-            } else if (!tl.line().text().isBlank()) {
-                // Include non-anchor lines (labels) only if they have text and are within
-                // proximity.
-                current.add(tl);
-            }
-        }
-
-        if (!current.isEmpty()) groups.add(current);
-
-        return groups.stream().filter(g -> hasEnoughAnchorRows(g, columnGrid)).toList();
-    }
-
-    private boolean hasEnoughAnchorRows(List<TokenizedLine> group, List<Float> columnGrid) {
-        return group.stream().filter(r -> r.isAnchor() && matchesGrid(r, columnGrid)).count()
-                >= MIN_TABLE_ROWS;
-    }
-
-    /** A line "matches" the grid when ≥ 60 % of its numeric tokens land in confirmed columns. */
-    private boolean matchesGrid(TokenizedLine tl, List<Float> columnGrid) {
-        if (tl.numeric().isEmpty()) return false;
-        long matches =
-                tl.numeric().stream()
-                        .filter(t -> nearestColumnIndex(t.right(), columnGrid) >= 0)
-                        .count();
-        return (double) matches / tl.numeric().size() >= 0.60;
-    }
-
-    private boolean hasInconsistentColumnMatch(TokenizedLine tl, List<Float> columnGrid) {
-        if (tl.numeric().isEmpty()) return false;
-        long hits =
-                tl.numeric().stream()
-                        .filter(t -> nearestColumnIndex(t.right(), columnGrid) >= 0)
-                        .count();
-        return (double) hits / tl.numeric().size() < 0.60;
-    }
-
-    // ── fragment assembly ────────────────────────────────────────────────────────────────────────
-
-    private Optional<TableFragment> buildFragment(
-            List<TokenizedLine> group, List<Float> columnGrid, int pageNumber, int tableIndex) {
-
-        long anchorCount =
-                group.stream().filter(r -> r.isAnchor() && matchesGrid(r, columnGrid)).count();
-        if (anchorCount < MIN_TABLE_ROWS) return Optional.empty();
-
-        List<String> warnings = new ArrayList<>();
-        List<List<String>> rawRows = new ArrayList<>();
-        List<TableRow> rows = new ArrayList<>();
-
-        for (int rowIdx = 0; rowIdx < group.size(); rowIdx++) {
-            TokenizedLine tl = group.get(rowIdx);
-            List<String> rawRow = buildRawRow(tl, columnGrid);
-            rawRows.add(Collections.unmodifiableList(rawRow));
-            rows.add(buildTableRow(rowIdx, tl, rawRow, columnGrid));
-        }
-
-        // Column count = 1 label column + confirmed numeric columns
-        int colCount = columnGrid.size() + 1;
-        Bounds bounds = computeGroupBounds(group);
-        float confidence = computeConfidence(group, columnGrid, warnings);
-
-        return Optional.of(
-                new TableFragment(
-                        "tbl-la-p" + pageNumber + "-" + tableIndex,
-                        pageNumber,
-                        bounds,
-                        List.of(),
-                        Collections.unmodifiableList(rows),
-                        Collections.unmodifiableList(rawRows),
-                        colCount,
-                        confidence,
-                        Collections.unmodifiableList(warnings),
-                        null));
-    }
-
-    /**
-     * Builds a raw row as a list of strings: index 0 = label text, indices 1..N = column values.
-     */
-    private List<String> buildRawRow(TokenizedLine tl, List<Float> columnGrid) {
-        String[] cells = new String[columnGrid.size() + 1];
-        Arrays.fill(cells, "");
-
-        // Separate label tokens (those not landing in any confirmed column) from column tokens.
-        List<String> labelParts = new ArrayList<>();
-        for (LineToken token : tl.all()) {
-            int col = nearestColumnIndex(token.right(), columnGrid);
-            if (col >= 0 && token.numeric()) {
-                int cellIdx = col + 1;
-                cells[cellIdx] =
-                        cells[cellIdx].isEmpty()
-                                ? token.text()
-                                : cells[cellIdx] + " " + token.text();
-            } else {
-                labelParts.add(token.text());
-            }
-        }
-        cells[0] = String.join(" ", labelParts).trim();
-        return Arrays.asList(cells);
-    }
-
-    private TableRow buildTableRow(
-            int rowIdx, TokenizedLine tl, List<String> rawRow, List<Float> columnGrid) {
-        List<TableCell> cells = new ArrayList<>(rawRow.size());
-
-        // Label cell: use the line's full bounds as an approximation.
-        cells.add(TableCell.of(0, rawRow.get(0), tl.line().bounds()));
-
-        for (int col = 0; col < columnGrid.size(); col++) {
-            String text = col + 1 < rawRow.size() ? rawRow.get(col + 1) : "";
-            float right = columnGrid.get(col);
-            float left = col > 0 ? columnGrid.get(col - 1) : right - 50f;
-            Bounds cellBounds =
-                    new Bounds(
-                            left,
-                            tl.line().bounds().y(),
-                            right - left,
-                            tl.line().bounds().height());
-            cells.add(TableCell.of(col + 1, text, cellBounds));
-        }
-        return new TableRow(rowIdx, Collections.unmodifiableList(cells));
-    }
-
-    // ── confidence scoring ───────────────────────────────────────────────────────────────────────
-
-    /**
-     * Heuristic score in [0.0, 0.85] (ceiling keeps results below Tabula lattice which starts at
-     * 1.0). Base 0.70; +0.05/col beyond 2 (max +0.10); +0.05 at ≥5 anchors, +0.05 at ≥8; −0.15 if
-     * &gt;30 % of anchors have inconsistent columns; −0.10 if non-anchors outnumber anchors.
-     */
-    private float computeConfidence(
-            List<TokenizedLine> group, List<Float> columnGrid, List<String> warnings) {
-        float score = 0.70f;
-
-        long anchorCount =
-                group.stream().filter(r -> r.isAnchor() && matchesGrid(r, columnGrid)).count();
-        long totalRows = group.size();
-
-        // More columns
-        int extraCols = Math.min(columnGrid.size() - MIN_COLUMNS, 2);
-        score += extraCols * 0.05f;
-
-        // More anchor rows
-        if (anchorCount >= 5) score += 0.05f;
-        if (anchorCount >= 8) score += 0.05f;
-
-        // Inconsistent column matching
-        long inconsistent =
-                group.stream()
-                        .filter(TokenizedLine::isAnchor)
-                        .filter(tl -> hasInconsistentColumnMatch(tl, columnGrid))
-                        .count();
-        if (inconsistent > anchorCount * 0.30) {
-            score -= 0.15f;
-            warnings.add(
-                    "Column match inconsistent on "
-                            + inconsistent
-                            + "/"
-                            + anchorCount
-                            + " anchor rows");
-        }
-
-        // Label-heavy
-        long nonAnchor = totalRows - anchorCount;
-        if (nonAnchor > anchorCount) {
-            score -= 0.10f;
-            warnings.add(
-                    "Non-anchor rows ("
-                            + nonAnchor
-                            + ") outnumber anchor rows ("
-                            + anchorCount
-                            + ")");
-        }
-
-        return Math.max(0f, Math.min(0.85f, score));
-    }
-
-    // ── utility ──────────────────────────────────────────────────────────────────────────────────
-
-    /**
-     * Returns the grid index nearest to {@code rightEdge}, or -1 if none is within {@value
-     * #COLUMN_MATCH_BUCKETS} buckets.
-     */
-    private int nearestColumnIndex(float rightEdge, List<Float> grid) {
-        int nearest = -1;
-        float minDist = COLUMN_MATCH_BUCKETS * COLUMN_BUCKET_PT + 1f;
-        for (int i = 0; i < grid.size(); i++) {
-            float dist = Math.abs(rightEdge - grid.get(i));
-            if (dist < minDist) {
-                minDist = dist;
-                nearest = i;
-            }
-        }
-        return nearest;
-    }
-
-    private Bounds computeGroupBounds(List<TokenizedLine> group) {
-        return group.stream()
-                .map(tl -> tl.line().bounds())
-                .reduce(Bounds::merge)
-                .orElse(new Bounds(0, 0, 0, 0));
-    }
-
-    /** Modal gap between consecutive line edges, used to calibrate the group-split threshold. */
-    private float computeModalSpacing(List<RawLine> lines) {
-        if (lines.size() < 2) return 0f;
-        Map<Float, Long> freq = new HashMap<>();
-        for (int i = 1; i < lines.size(); i++) {
-            float gap = lines.get(i).bounds().y() - lines.get(i - 1).bounds().bottom();
-            if (gap > 0) freq.merge(Math.round(gap / 2f) * 2f, 1L, Long::sum);
-        }
-        return freq.entrySet().stream()
-                .max(Map.Entry.comparingByValue())
-                .map(Map.Entry::getKey)
-                .orElse(0f);
-    }
-
-    private static int bucket(float x) {
-        return Math.round(x / COLUMN_BUCKET_PT);
-    }
-
-    // ── private data types ───────────────────────────────────────────────────────────────────────
-
-    /** A word-level token with an approximate right-edge x-position. */
-    record LineToken(String text, float x, float right, boolean numeric) {}
-
-    /** A {@link RawLine} with tokens pre-computed; an "anchor" has ≥ 2 numeric tokens. */
-    record TokenizedLine(RawLine line, List<LineToken> all, List<LineToken> numeric) {
-        boolean isAnchor() {
-            return numeric.size() >= 2;
-        }
-    }
-}

app/common/src/main/java/stirling/software/SPDF/pdf/parser/LineBuilder.java

@@ -1,139 +0,0 @@
-package stirling.software.SPDF.pdf.parser;
-
-import static stirling.software.SPDF.pdf.parser.PdfModels.*;
-
-import java.util.ArrayList;
-import java.util.Comparator;
-import java.util.List;
-
-import org.springframework.stereotype.Service;
-
-import lombok.extern.slf4j.Slf4j;
-
-/**
- * Groups {@link TextFragment} objects into visual {@link RawLine}s using baseline proximity.
- *
- * <p>Fragments are on the same line when their baselines are within a font-size-derived tolerance.
- * A new line starts whenever the horizontal gap exceeds an adaptive column-gap threshold ({@code
- * max(effectiveWidth * COLUMN_GAP_RATIO, COLUMN_GAP_MIN_PT)}), splitting two-column text.
- */
-@Service
-@Slf4j
-public class LineBuilder {
-
-    /** Baseline tolerance as a fraction of font size; 0.5 keeps mixed-size text on one line. */
-    private static final float BASELINE_TOLERANCE_FACTOR = 0.5f;
-
-    /** Absolute minimum tolerance so tiny font sizes don't collapse multi-line content. */
-    private static final float MIN_BASELINE_TOLERANCE = 2f;
-
-    /**
-     * Column-gap threshold as a fraction of page width; 0.10 clears tab stops but stays below
-     * two-column gutters.
-     */
-    static final float COLUMN_GAP_RATIO = 0.10f;
-
-    /** Floor for the column-gap threshold so narrow pages don't over-split lines. */
-    static final float COLUMN_GAP_MIN_PT = 40f;
-
-    public List<RawLine> build(List<TextFragment> fragments, int pageNumber) {
-        if (fragments.isEmpty()) return List.of();
-
-        float effectiveWidth = inferEffectiveWidth(fragments);
-        float columnGapThreshold = Math.max(effectiveWidth * COLUMN_GAP_RATIO, COLUMN_GAP_MIN_PT);
-        log.debug(
-                "LineBuilder page {}: effectiveWidth={:.1f}pt, columnGapThreshold={:.1f}pt",
-                pageNumber,
-                effectiveWidth,
-                columnGapThreshold);
-
-        // Sort top-to-bottom first, then left-to-right within the same baseline band.
-        List<TextFragment> sorted =
-                fragments.stream()
-                        .sorted(
-                                Comparator.comparingDouble(TextFragment::baseline)
-                                        .thenComparingDouble(f -> f.bounds().x()))
-                        .toList();
-
-        List<List<TextFragment>> groups = groupByBaseline(sorted, columnGapThreshold);
-
-        List<RawLine> lines = new ArrayList<>(groups.size());
-        for (int i = 0; i < groups.size(); i++) {
-            List<TextFragment> group =
-                    groups.get(i).stream()
-                            .sorted(Comparator.comparingDouble(f -> f.bounds().x()))
-                            .toList();
-
-            Bounds lineBounds =
-                    group.stream()
-                            .map(TextFragment::bounds)
-                            .reduce(Bounds::merge)
-                            .orElse(new Bounds(0, 0, 0, 0));
-
-            lines.add(new RawLine("ln-p" + pageNumber + "-" + i, group, lineBounds, pageNumber));
-        }
-        return lines;
-    }
-
-    private List<List<TextFragment>> groupByBaseline(
-            List<TextFragment> sorted, float columnGapThreshold) {
-        List<List<TextFragment>> groups = new ArrayList<>();
-        List<TextFragment> current = new ArrayList<>();
-        float currentBaseline = Float.NaN;
-
-        for (TextFragment fragment : sorted) {
-            if (current.isEmpty()) {
-                current.add(fragment);
-                currentBaseline = fragment.baseline();
-                continue;
-            }
-
-            float maxFontSize =
-                    Math.max(
-                            fragment.fontSize(),
-                            (float)
-                                    current.stream()
-                                            .mapToDouble(TextFragment::fontSize)
-                                            .max()
-                                            .orElse(0));
-            float tolerance =
-                    Math.max(maxFontSize * BASELINE_TOLERANCE_FACTOR, MIN_BASELINE_TOLERANCE);
-
-            boolean sameBaseline = Math.abs(fragment.baseline() - currentBaseline) <= tolerance;
-            boolean columnGap = sameBaseline && hasColumnGap(fragment, current, columnGapThreshold);
-
-            if (sameBaseline && !columnGap) {
-                current.add(fragment);
-                // Anchor to the weighted mean baseline so long lines stay stable.
-                currentBaseline =
-                        (currentBaseline * (current.size() - 1) + fragment.baseline())
-                                / current.size();
-            } else {
-                groups.add(current);
-                current = new ArrayList<>();
-                current.add(fragment);
-                currentBaseline = fragment.baseline();
-            }
-        }
-
-        if (!current.isEmpty()) groups.add(current);
-        return groups;
-    }
-
-    /**
-     * True when the gap from the rightmost fragment in {@code group} to {@code next} exceeds {@code
-     * threshold}.
-     */
-    private static boolean hasColumnGap(
-            TextFragment next, List<TextFragment> group, float threshold) {
-        float lastRight = group.get(group.size() - 1).bounds().right();
-        return next.bounds().x() - lastRight > threshold;
-    }
-
-    /** Infers effective page width from the rightmost fragment right-edge plus a 10 % margin. */
-    private static float inferEffectiveWidth(List<TextFragment> fragments) {
-        double maxRight =
-                fragments.stream().mapToDouble(f -> f.bounds().right()).max().orElse(500.0);
-        return (float) maxRight * 1.10f;
-    }
-}

app/common/src/main/java/stirling/software/SPDF/pdf/parser/PdfIngester.java

@@ -1,79 +0,0 @@
-package stirling.software.SPDF.pdf.parser;
-
-import static stirling.software.SPDF.pdf.parser.PdfModels.*;
-
-import java.io.IOException;
-import java.util.ArrayList;
-import java.util.List;
-
-import org.apache.pdfbox.pdmodel.PDDocument;
-import org.apache.pdfbox.pdmodel.PDPage;
-import org.apache.pdfbox.pdmodel.common.PDRectangle;
-import org.springframework.stereotype.Service;
-
-import lombok.RequiredArgsConstructor;
-import lombok.extern.slf4j.Slf4j;
-
-/**
- * Runs the per-page ingestion pipeline: {@link WordExtractingStripper} → {@link LineBuilder} →
- * {@link TableParser}, producing a {@link PdfModels.ParsedPage} per page. The caller owns the
- * {@link PDDocument} lifecycle.
- */
-@Service
-@RequiredArgsConstructor
-@Slf4j
-public class PdfIngester {
-
-    private final LineBuilder lineBuilder;
-    private final TableParser tableParser;
-
-    public List<ParsedPage> parse(PDDocument document) throws IOException {
-        return parse(document, document.getNumberOfPages());
-    }
-
-    public List<ParsedPage> parse(PDDocument document, int maxPages) throws IOException {
-        int pageCount = Math.min(document.getNumberOfPages(), maxPages);
-        List<ParsedPage> pages = new ArrayList<>(pageCount);
-        long fragmentsMs = 0;
-        long tablesMs = 0;
-        long t0 = System.currentTimeMillis();
-
-        for (int p = 1; p <= pageCount; p++) {
-            long ft = System.currentTimeMillis();
-            List<TextFragment> fragments = extractFragments(document, p);
-            fragmentsMs += System.currentTimeMillis() - ft;
-
-            PDPage page = document.getPage(p - 1);
-            PDRectangle mediaBox = page.getMediaBox();
-            List<RawLine> lines = lineBuilder.build(fragments, p);
-            RawPage rawPage = new RawPage(p, mediaBox.getWidth(), mediaBox.getHeight(), lines);
-
-            long tt = System.currentTimeMillis();
-            List<TableFragment> tables = tableParser.parse(document, rawPage);
-            tablesMs += System.currentTimeMillis() - tt;
-
-            log.debug(
-                    "Page {}: {} fragments → {} lines, {} table(s)",
-                    p,
-                    fragments.size(),
-                    lines.size(),
-                    tables.size());
-            pages.add(new ParsedPage(p, mediaBox.getWidth(), mediaBox.getHeight(), tables, lines));
-        }
-
-        log.info(
-                "[timing] parse pages={} total={}ms fragments={}ms tables={}ms",
-                pageCount,
-                System.currentTimeMillis() - t0,
-                fragmentsMs,
-                tablesMs);
-        return pages;
-    }
-
-    private List<TextFragment> extractFragments(PDDocument document, int pageNumber)
-            throws IOException {
-        WordExtractingStripper stripper = new WordExtractingStripper(pageNumber);
-        stripper.getText(document);
-        return stripper.getFragments();
-    }
-}

app/common/src/main/java/stirling/software/SPDF/pdf/parser/WordExtractingStripper.java

@@ -1,113 +0,0 @@
-package stirling.software.SPDF.pdf.parser;
-
-import static stirling.software.SPDF.pdf.parser.PdfModels.*;
-
-import java.io.IOException;
-import java.util.ArrayList;
-import java.util.Collections;
-import java.util.List;
-
-import org.apache.pdfbox.pdmodel.PDPage;
-import org.apache.pdfbox.pdmodel.font.PDFont;
-import org.apache.pdfbox.text.PDFTextStripper;
-import org.apache.pdfbox.text.TextPosition;
-
-/**
- * Extends {@link PDFTextStripper} to capture per-fragment geometry and font metadata.
- *
- * <p>Overrides {@link #writeString} to split each content-stream string into word-level {@link
- * TextFragment}s with bounding boxes, baseline, font name, and bold flag. Coordinates are in
- * PDFTextStripper space: (0,0) top-left, Y increases downward, {@code getY()} is the baseline.
- */
-class WordExtractingStripper extends PDFTextStripper {
-
-    private final int targetPage;
-    private final List<TextFragment> fragments = new ArrayList<>();
-    private int fragmentIndex = 0;
-
-    WordExtractingStripper(int pageNumber) throws IOException {
-        this.targetPage = pageNumber;
-        setStartPage(pageNumber);
-        setEndPage(pageNumber);
-        setSortByPosition(true);
-    }
-
-    @Override
-    protected void startPage(PDPage page) throws IOException {
-        super.startPage(page);
-        fragments.clear();
-        fragmentIndex = 0;
-    }
-
-    @Override
-    protected void writeString(String text, List<TextPosition> textPositions) throws IOException {
-        if (text == null || text.isBlank()) return;
-
-        // Fast path: no whitespace → emit one fragment (most financial PDFs have each
-        // number as its own string operation, so this is the common case).
-        if (text.indexOf(' ') < 0) {
-            emitFragment(text, textPositions);
-            return;
-        }
-
-        // Per-word splitting requires 1:1 text-char to TextPosition correspondence.
-        // Fall back to one fragment when sizes differ (ligatures, encoding edge cases).
-        if (textPositions.size() != text.length()) {
-            emitFragment(text, textPositions);
-            return;
-        }
-
-        // Emit one TextFragment per whitespace-delimited word with accurate per-word bounds.
-        int start = 0;
-        for (int i = 0; i <= text.length(); i++) {
-            if (i == text.length() || text.charAt(i) == ' ') {
-                if (start < i) {
-                    emitFragment(text.substring(start, i), textPositions.subList(start, i));
-                }
-                start = i + 1;
-            }
-        }
-    }
-
-    private void emitFragment(String text, List<TextPosition> positions) {
-        if (positions.isEmpty()) return;
-
-        float minX = Float.MAX_VALUE;
-        float minY = Float.MAX_VALUE;
-        float maxRight = -Float.MAX_VALUE;
-        float maxBaseline = -Float.MAX_VALUE;
-        TextPosition first = null;
-
-        for (TextPosition tp : positions) {
-            if (tp == null) continue;
-            if (first == null) first = tp;
-
-            float x = tp.getX();
-            // getY() is the baseline; top of character = getY() - getHeight().
-            float top = tp.getY() - tp.getHeight();
-            float right = x + tp.getWidth();
-            float baseline = tp.getY();
-
-            minX = Math.min(minX, x);
-            minY = Math.min(minY, top);
-            maxRight = Math.max(maxRight, right);
-            maxBaseline = Math.max(maxBaseline, baseline);
-        }
-
-        if (first == null) return;
-
-        PDFont font = first.getFont();
-        String fontName = font != null ? font.getName() : "";
-        boolean bold = fontName != null && fontName.toLowerCase().contains("bold");
-        // getHeight() gives the rendered glyph height, which is the most reliable visual size.
-        float fontSize = first.getHeight();
-
-        Bounds bounds = new Bounds(minX, minY, maxRight - minX, maxBaseline - minY);
-        String id = "tf-p" + targetPage + "-" + fragmentIndex++;
-        fragments.add(new TextFragment(id, text, bounds, maxBaseline, fontSize, fontName, bold));
-    }
-
-    List<TextFragment> getFragments() {
-        return Collections.unmodifiableList(fragments);
-    }
-}

app/common/src/main/java/stirling/software/common/cluster/FileStore.java

@@ -11,23 +11,39 @@ public interface FileStore {
     /** Stored file record. */
     record Stored(String fileId, long size) {}
 
-    /** Store the given stream and return a generated file id and total bytes written. */
-    Stored store(InputStream in, String originalName) throws IOException;
+    /**
+     * Store the given stream and return a generated file id and total bytes written. {@code owner}
+     * may be null to indicate the file has no associated user (anonymous / desktop / async job with
+     * no propagated security context); a non-null value is persisted alongside the data so {@link
+     * #getOwner(String)} can return it later for authorization checks.
+     */
+    Stored store(InputStream in, String originalName, String owner) throws IOException;
+
+    /** Store with no owner. Equivalent to {@link #store(InputStream, String, String)} with null. */
+    default Stored store(InputStream in, String originalName) throws IOException {
+        return store(in, originalName, null);
+    }
 
     /**
      * Store the file at {@code source} and return a generated file id and total bytes written.
      *
      * <p>Default implementation opens {@code source} as a stream and delegates to {@link
-     * #store(InputStream, String)}. Local-disk implementations should override to use a direct
-     * file-to-file copy ({@code Files.copy(source, dest)} can use {@code sendfile(2)} on Linux),
-     * which avoids the two-memory-copy hit of streaming a disk-backed upload through the JVM heap.
+     * #store(InputStream, String, String)}. Local-disk implementations should override to use a
+     * direct file-to-file copy ({@code Files.copy(source, dest)} can use {@code sendfile(2)} on
+     * Linux), which avoids the two-memory-copy hit of streaming a disk-backed upload through the
+     * JVM heap.
      */
-    default Stored store(Path source, String originalName) throws IOException {
+    default Stored store(Path source, String originalName, String owner) throws IOException {
         try (InputStream in = Files.newInputStream(source)) {
-            return store(in, originalName);
+            return store(in, originalName, owner);
         }
     }
 
+    /** Store with no owner. Equivalent to {@link #store(Path, String, String)} with null. */
+    default Stored store(Path source, String originalName) throws IOException {
+        return store(source, originalName, null);
+    }
+
     /** Open the stored file for streaming reads. Caller closes. */
     InputStream retrieve(String fileId) throws IOException;
 
@@ -42,4 +58,12 @@ public interface FileStore {
 
     /** Whether the file id exists in the store. */
     boolean exists(String fileId);
+
+    /**
+     * Returns the owner identifier recorded at store time, or {@code null} if the file does not
+     * exist or was stored without an owner. Implementations must not throw when the file is missing
+     * or when the owner record is absent; they should return null so callers can treat "no owner"
+     * as a non-authoritative case.
+     */
+    String getOwner(String fileId) throws IOException;
 }

app/common/src/main/java/stirling/software/common/cluster/inprocess/LocalDiskFileStore.java

@@ -3,9 +3,12 @@ package stirling.software.common.cluster.inprocess;
 import java.io.BufferedInputStream;
 import java.io.IOException;
 import java.io.InputStream;
+import java.nio.charset.StandardCharsets;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.util.UUID;
+import java.util.concurrent.locks.ReentrantLock;
+import java.util.regex.Pattern;
 
 import lombok.extern.slf4j.Slf4j;
 
@@ -15,33 +18,47 @@ import stirling.software.common.cluster.FileStore;
 @Slf4j
 public class LocalDiskFileStore implements FileStore {
 
+    private static final String OWNER_SUFFIX = ".owner";
+
+    // File ids are generated as random UUIDs; reject anything else so a tainted id can never reach
+    // Files.* APIs (defence in depth on top of the resolve() prefix check, and silences CodeQL's
+    // path-injection finding on the resolveOwner sidecar lookup).
+    private static final Pattern UUID_PATTERN =
+            Pattern.compile(
+                    "^[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}$");
+
     private final String baseDirPath;
+    // Fixed-size lock stripes so concurrent store/delete on the same (or colliding) fileId
+    // serialise the data-file + owner-sidecar pair as one critical section. Striped (not
+    // per-id) so the map never has to be cleaned up; collisions across unrelated ids are
+    // harmless contention.
+    private static final int LOCK_STRIPES = 64;
+    private final ReentrantLock[] stripes = new ReentrantLock[LOCK_STRIPES];
 
     public LocalDiskFileStore(String baseDirPath) {
         this.baseDirPath = baseDirPath;
+        for (int i = 0; i < LOCK_STRIPES; i++) {
+            stripes[i] = new ReentrantLock();
+        }
     }
 
     @Override
-    public Stored store(InputStream in, String originalName) throws IOException {
+    public Stored store(InputStream in, String originalName, String owner) throws IOException {
         String fileId = UUID.randomUUID().toString();
         Path filePath = resolve(fileId);
         Files.createDirectories(filePath.getParent());
+        ReentrantLock lock = acquire(fileId);
         boolean success = false;
         try {
             long size = Files.copy(in, filePath);
+            writeOwner(fileId, owner);
             success = true;
             return new Stored(fileId, size);
         } finally {
             if (!success) {
-                try {
-                    Files.deleteIfExists(filePath);
-                } catch (IOException cleanupEx) {
-                    log.warn(
-                            "Failed to clean up partial file {} after store failure",
-                            filePath,
-                            cleanupEx);
-                }
+                cleanupAfterFailedStore(fileId, filePath);
             }
+            release(fileId, lock);
         }
     }
 
@@ -52,27 +69,44 @@ public class LocalDiskFileStore implements FileStore {
      * the source size before copying so the post-copy stat is unnecessary.
      */
     @Override
-    public Stored store(Path source, String originalName) throws IOException {
+    public Stored store(Path source, String originalName, String owner) throws IOException {
         String fileId = UUID.randomUUID().toString();
         Path filePath = resolve(fileId);
         Files.createDirectories(filePath.getParent());
         long size = Files.size(source);
+        ReentrantLock lock = acquire(fileId);
         boolean success = false;
         try {
             Files.copy(source, filePath);
+            writeOwner(fileId, owner);
             success = true;
             return new Stored(fileId, size);
         } finally {
             if (!success) {
-                try {
-                    Files.deleteIfExists(filePath);
-                } catch (IOException cleanupEx) {
-                    log.warn(
-                            "Failed to clean up partial file {} after store failure",
-                            filePath,
-                            cleanupEx);
-                }
+                cleanupAfterFailedStore(fileId, filePath);
             }
+            release(fileId, lock);
+        }
+    }
+
+    private void writeOwner(String fileId, String owner) throws IOException {
+        if (owner == null || owner.isBlank()) {
+            return;
+        }
+        Path ownerPath = resolveOwner(fileId);
+        Files.write(ownerPath, owner.getBytes(StandardCharsets.UTF_8));
+    }
+
+    private void cleanupAfterFailedStore(String fileId, Path filePath) {
+        try {
+            Files.deleteIfExists(filePath);
+        } catch (IOException cleanupEx) {
+            log.warn("Failed to clean up partial file {} after store failure", filePath, cleanupEx);
+        }
+        try {
+            Files.deleteIfExists(resolveOwner(fileId));
+        } catch (IOException cleanupEx) {
+            log.warn("Failed to clean up owner sidecar for {} after store failure", fileId);
         }
     }
 
@@ -101,11 +135,26 @@ public class LocalDiskFileStore implements FileStore {
 
     @Override
     public boolean delete(String fileId) {
+        ReentrantLock lock = acquire(fileId);
         try {
-            return Files.deleteIfExists(resolve(fileId));
-        } catch (IOException e) {
-            log.error("Error deleting file with ID: {}", fileId, e);
-            return false;
+            // Data first, owner second: a concurrent retrieve that observes the transient
+            // (data-gone, owner-still-present) window simply fails with IOException; the inverse
+            // order would briefly look like an unowned file and could grant cross-user access.
+            boolean removed;
+            try {
+                removed = Files.deleteIfExists(resolve(fileId));
+            } catch (IOException e) {
+                log.error("Error deleting file with ID: {}", fileId, e);
+                return false;
+            }
+            try {
+                Files.deleteIfExists(resolveOwner(fileId));
+            } catch (IOException e) {
+                log.warn("Error deleting owner sidecar for file ID: {}", fileId, e);
+            }
+            return removed;
+        } finally {
+            release(fileId, lock);
         }
     }
 
@@ -114,8 +163,21 @@ public class LocalDiskFileStore implements FileStore {
         return Files.exists(resolve(fileId));
     }
 
+    @Override
+    public String getOwner(String fileId) throws IOException {
+        Path ownerPath = resolveOwner(fileId);
+        if (!Files.exists(ownerPath)) {
+            return null;
+        }
+        byte[] bytes = Files.readAllBytes(ownerPath);
+        if (bytes.length == 0) {
+            return null;
+        }
+        return new String(bytes, StandardCharsets.UTF_8);
+    }
+
     public Path resolve(String fileId) {
-        if (fileId.contains("..") || fileId.contains("/") || fileId.contains("\\")) {
+        if (fileId == null || !UUID_PATTERN.matcher(fileId).matches()) {
             throw new IllegalArgumentException("Invalid file ID");
         }
         Path basePath = Path.of(baseDirPath).normalize().toAbsolutePath();
@@ -125,4 +187,19 @@ public class LocalDiskFileStore implements FileStore {
         }
         return resolvedPath;
     }
+
+    private Path resolveOwner(String fileId) {
+        Path data = resolve(fileId);
+        return data.resolveSibling(data.getFileName().toString() + OWNER_SUFFIX);
+    }
+
+    private ReentrantLock acquire(String fileId) {
+        ReentrantLock lock = stripes[(fileId.hashCode() & Integer.MAX_VALUE) % LOCK_STRIPES];
+        lock.lock();
+        return lock;
+    }
+
+    private void release(String fileId, ReentrantLock lock) {
+        lock.unlock();
+    }
 }

app/common/src/main/java/stirling/software/common/model/ApplicationProperties.java

@@ -77,8 +77,10 @@ public class ApplicationProperties {
     private ProcessExecutor processExecutor = new ProcessExecutor();
     private PdfEditor pdfEditor = new PdfEditor();
     private AiEngine aiEngine = new AiEngine();
+    private Mcp mcp = new Mcp();
     private InternalApi internalApi = new InternalApi();
     private Cluster cluster = new Cluster();
+    private Policies policies = new Policies();
 
     @Bean
     public PropertySource<?> dynamicYamlPropertySource(ConfigurableEnvironment environment)
@@ -202,6 +204,45 @@ public class ApplicationProperties {
         }
     }
 
+    @Data
+    public static class Policies {
+        /**
+         * Absolute directories that policy folder input sources and output sinks may read from or
+         * write to. Empty (the default) disables folder access entirely, so a policy can never be
+         * pointed at an arbitrary server path. Stirling's own config directory is always
+         * off-limits, and folder access is always disabled in SaaS mode regardless of this list.
+         */
+        private List<String> allowedFolderRoots = new java.util.ArrayList<>();
+
+        /** How often (seconds) the schedule trigger checks for policies whose schedule is due. */
+        private long scheduleSweepSeconds = 60;
+
+        /**
+         * How often (seconds) the folder-watch trigger reconciles its watch registrations and
+         * re-runs every folder-watch policy as a safety net for filesystem events that were missed
+         * (NFS, bind mounts, inotify-queue overflow).
+         */
+        private long watchReconcileSeconds = 300;
+
+        /**
+         * How long (milliseconds) the folder-watch trigger keeps draining filesystem events after
+         * the first, so a burst from a single file copy coalesces into one run instead of many.
+         */
+        private long watchQuietPeriodMs = 500;
+
+        /**
+         * SSE emitter timeout (milliseconds) for streamed runs; generous for long multi-step runs.
+         */
+        private long streamTimeoutMs = 1800000;
+
+        /**
+         * How long (minutes) a finished run's in-memory state is retained before eviction,
+         * mirroring the job-result expiry so rich run state does not outlive the process. Active
+         * and paused runs are kept regardless of age.
+         */
+        private int runExpiryMinutes = 30;
+    }
+
     @Data
     public static class PdfEditor {
         private Cache cache = new Cache();
@@ -256,6 +297,94 @@ public class ApplicationProperties {
         private int longRunningTimeoutSeconds = 600;
     }
 
+    /**
+     * Model Context Protocol (MCP) server configuration. All keys live under the top-level {@code
+     * mcp.*} prefix. {@link #enabled} defaults to {@code false}: when off, no MCP beans are wired,
+     * no /mcp endpoint exists, and no protected-resource metadata is published.
+     */
+    @Data
+    public static class Mcp {
+
+        /** Master switch. When {@code false} (default), no MCP beans are wired. */
+        private boolean enabled = false;
+
+        /**
+         * When {@code true} (default), invocations require an OAuth scope: {@code mcp.tools.read}
+         * for read-style operations and {@code mcp.tools.write} for write/destructive ones. When
+         * {@code false}, scope checks are skipped (use only if your IdP issues a single coarse
+         * scope).
+         */
+        private boolean scopesEnabled = true;
+
+        /** How often to refresh the AI capabilities manifest from the engine. */
+        private int engineCapabilityRefreshMinutes = 5;
+
+        /**
+         * Tool allow-list (operation ids, e.g. {@code compress-pdf}). When non-empty, ONLY these
+         * operations are exposed over MCP; everything else is hidden, undescribable, and
+         * uninvocable - on top of the global endpoint enable/disable config. Empty = allow all.
+         */
+        private List<String> allowedOperations = new ArrayList<>();
+
+        /**
+         * Tool deny-list (operation ids). Any operation listed here is removed from MCP even if it
+         * would otherwise be allowed. Applied after {@link #allowedOperations}.
+         */
+        private List<String> blockedOperations = new ArrayList<>();
+
+        /** Max MCP request body size in bytes; inline file uploads ride in the JSON-RPC body. */
+        private long maxRequestBytes = 10L * 1024 * 1024;
+
+        /** Results up to this size return inline as base64; larger ones return a fileId only. */
+        private long maxInlineResponseBytes = 10L * 1024 * 1024;
+
+        private Auth auth = new Auth();
+
+        @Data
+        public static class Auth {
+            /**
+             * Authentication mode for the MCP endpoint. {@code oauth} (default) runs a full OAuth2
+             * resource server (JWT, RFC 8707 audience, RFC 9728 metadata). {@code apikey} accepts a
+             * Stirling per-user API key via the {@code X-API-KEY} header (or {@code Authorization:
+             * Bearer <key>}) and binds the request to that user - the low-friction self-host path,
+             * no external IdP required.
+             */
+            private String mode = "oauth";
+
+            /** OAuth2 issuer URI, e.g. {@code http://localhost:9000}. Required when MCP is on. */
+            private String issuerUri = "";
+
+            /**
+             * JWKS URI. When blank, derived from the issuer's {@code
+             * /.well-known/openid-configuration} document.
+             */
+            private String jwksUri = "";
+
+            /**
+             * RFC 8707 resource identifier of THIS MCP server, e.g. {@code
+             * http://localhost:8080/mcp}. Tokens that do not list this id in their {@code aud}
+             * claim are rejected with HTTP 401.
+             */
+            private String resourceId = "";
+
+            /**
+             * JWT claim whose value is matched against a provisioned Stirling username. Defaults to
+             * {@code sub}; set to {@code email} or {@code preferred_username} to match how your IdP
+             * maps users to Stirling accounts.
+             */
+            private String usernameClaim = "sub";
+
+            /**
+             * When {@code true} (default), a validated token is accepted only if its {@link
+             * #usernameClaim} value resolves to an existing, enabled Stirling user account. Tokens
+             * whose subject has no Stirling account (or a disabled one) are rejected with HTTP 403.
+             * Set to {@code false} only if you intentionally want any IdP-valid token to use MCP
+             * without a local account.
+             */
+            private boolean requireExistingAccount = true;
+        }
+    }
+
     /**
      * Cluster backplane configuration. All keys live under the top-level {@code cluster.*} prefix
      * (e.g. env var {@code CLUSTER_ENABLED}). The master switch is {@link #enabled} and defaults to

app/common/src/main/java/stirling/software/common/pdf/HeadingDetector.java

@@ -0,0 +1,191 @@
+package stirling.software.common.pdf;
+
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import stirling.software.jpdfium.text.PageText;
+import stirling.software.jpdfium.text.TextChar;
+import stirling.software.jpdfium.text.TextLine;
+import stirling.software.jpdfium.text.TextWord;
+
+final class HeadingDetector {
+
+    private HeadingDetector() {}
+
+    /** A heading is at most this many words; longer lines are treated as body text. */
+    private static final int MAX_HEADING_WORDS = 12;
+
+    /**
+     * Returns the Markdown heading prefix for a line. The decision combines several signals, never
+     * text matching, so a plain line that merely shares text with a heading is never promoted:
+     *
+     * <ul>
+     *   <li><b>Size</b> — dominant glyph font size vs. the document body median (primary signal).
+     *       Some PDFs encode visual size in the text matrix, so every glyph reports ~1.0; for those
+     *       the line height is used as the proxy instead.
+     *   <li><b>Brevity</b> — headings are short labels; a line over {@value #MAX_HEADING_WORDS}
+     *       words is body text regardless of size.
+     *   <li><b>Not a sentence</b> — a line ending in {@code . ! ?} reads as prose, not a heading.
+     * </ul>
+     *
+     * <p>Boldness is deliberately <em>not</em> a heading signal — a bold-but-not-larger line is
+     * emphasis, not a heading (see {@link #isBoldLabel}); promoting it to {@code #}/{@code ##} is
+     * the main source of false-positive headings.
+     *
+     * <ul>
+     *   <li>size &gt; baseline * 1.4 → {@code "# "}
+     *   <li>size &gt; baseline * 1.2 → {@code "## "}
+     *   <li>otherwise → {@code ""}
+     * </ul>
+     */
+    static String headingPrefix(TextLine line, float medianBodySize, float medianBodyHeight) {
+        String text = line.text().strip();
+        if (text.isEmpty() || wordCount(text) > MAX_HEADING_WORDS || endsLikeSentence(text)) {
+            return "";
+        }
+
+        float dominant = dominantFontSize(line);
+        float value;
+        float baseline;
+        if (dominant > 2f && medianBodySize > 2f) {
+            value = dominant;
+            baseline = medianBodySize;
+        } else {
+            value = line.height();
+            baseline = medianBodyHeight;
+        }
+        if (baseline <= 0f) {
+            return "";
+        }
+
+        float ratio = value / baseline;
+        if (ratio > 1.4f) {
+            return "# ";
+        }
+        if (ratio > 1.2f) {
+            return "## ";
+        }
+        return "";
+    }
+
+    /**
+     * True when a line should be emphasised as bold (rendered {@code **like this**}) rather than
+     * promoted to a heading: it is bold, short, and not a full sentence. Used for bold labels that
+     * are not large enough to be headings.
+     */
+    static boolean isBoldLabel(TextLine line) {
+        String text = line.text().strip();
+        if (text.isEmpty() || wordCount(text) > MAX_HEADING_WORDS || endsLikeSentence(text)) {
+            return false;
+        }
+        return isBold(line);
+    }
+
+    private static int wordCount(String text) {
+        return text.split("\\s+").length;
+    }
+
+    private static boolean endsLikeSentence(String text) {
+        char last = text.charAt(text.length() - 1);
+        return last == '.' || last == '!' || last == '?';
+    }
+
+    /** True when the line's dominant font is bold, inferred from PostScript font names. */
+    private static boolean isBold(TextLine line) {
+        Map<String, Integer> counts = new HashMap<>();
+        for (TextWord word : line.words()) {
+            for (TextChar ch : word.chars()) {
+                if (ch.isWhitespace() || ch.isNewline()) {
+                    continue;
+                }
+                String name = ch.fontName();
+                if (name != null && !name.isBlank()) {
+                    counts.merge(name, 1, Integer::sum);
+                }
+            }
+        }
+        String dominantFont = "";
+        int max = -1;
+        for (Map.Entry<String, Integer> e : counts.entrySet()) {
+            if (e.getValue() > max) {
+                max = e.getValue();
+                dominantFont = e.getKey();
+            }
+        }
+        String lower = dominantFont.toLowerCase(java.util.Locale.ROOT);
+        return lower.contains("bold")
+                || lower.contains("black")
+                || lower.contains("heavy")
+                || lower.contains("semibold");
+    }
+
+    /** Computes the median glyph font size across all pages. */
+    static float medianFontSize(List<PageText> allPages) {
+        List<Float> sizes = new ArrayList<>();
+        for (PageText page : allPages) {
+            for (TextChar ch : page.chars()) {
+                if (!ch.isWhitespace() && !ch.isNewline() && ch.fontSize() > 0f) {
+                    sizes.add(ch.fontSize());
+                }
+            }
+        }
+        return median(sizes, 12f);
+    }
+
+    /** Computes the median TextLine height across all pages. Used when font size is degenerate. */
+    static float medianLineHeight(List<PageText> allPages) {
+        List<Float> heights = new ArrayList<>();
+        for (PageText page : allPages) {
+            for (TextLine line : page.lines()) {
+                if (line.height() > 0f && !line.text().isBlank()) {
+                    heights.add(line.height());
+                }
+            }
+        }
+        return median(heights, 12f);
+    }
+
+    private static float median(List<Float> values, float fallback) {
+        if (values.isEmpty()) {
+            return fallback;
+        }
+        Collections.sort(values);
+        int mid = values.size() / 2;
+        if (values.size() % 2 == 0) {
+            return (values.get(mid - 1) + values.get(mid)) / 2f;
+        }
+        return values.get(mid);
+    }
+
+    /**
+     * Returns the font size that appears most often (by character count) in the given line. Ties
+     * are broken in favour of the larger size.
+     */
+    private static float dominantFontSize(TextLine line) {
+        Map<Float, Integer> counts = new HashMap<>();
+        for (TextWord word : line.words()) {
+            for (TextChar ch : word.chars()) {
+                if (!ch.isWhitespace() && !ch.isNewline() && ch.fontSize() > 0f) {
+                    counts.merge(ch.fontSize(), 1, Integer::sum);
+                }
+            }
+        }
+        if (counts.isEmpty()) {
+            return 0f;
+        }
+        float dominant = 0f;
+        int maxCount = -1;
+        for (Map.Entry<Float, Integer> entry : counts.entrySet()) {
+            int count = entry.getValue();
+            float size = entry.getKey();
+            if (count > maxCount || (count == maxCount && size > dominant)) {
+                maxCount = count;
+                dominant = size;
+            }
+        }
+        return dominant;
+    }
+}

app/common/src/main/java/stirling/software/common/pdf/TableRenderer.java

@@ -0,0 +1,82 @@
+package stirling.software.common.pdf;
+
+import stirling.software.jpdfium.text.Table;
+
+final class TableRenderer {
+    private TableRenderer() {}
+
+    /** Renders a Table as a GitHub-Flavoured Markdown table string. */
+    static String render(Table table) {
+        if (table.rowCount() == 0) {
+            return "";
+        }
+
+        String[][] grid = table.asGrid();
+
+        if (table.rowCount() < 2) {
+            // No separator row possible — return plain lines
+            StringBuilder sb = new StringBuilder();
+            for (int c = 0; c < grid[0].length; c++) {
+                if (c > 0) sb.append('\n');
+                sb.append(escape(grid[0][c].trim()));
+            }
+            return sb.toString();
+        }
+
+        int cols = grid[0].length;
+
+        // Compute column widths: max(3, max content length across all rows)
+        int[] widths = new int[cols];
+        for (int c = 0; c < cols; c++) {
+            widths[c] = 3;
+        }
+        for (String[] row : grid) {
+            for (int c = 0; c < cols; c++) {
+                String cell = c < row.length ? row[c].trim() : "";
+                widths[c] = Math.max(widths[c], escape(cell).length());
+            }
+        }
+
+        StringBuilder sb = new StringBuilder();
+
+        // Header row
+        sb.append(buildRow(grid[0], widths, cols));
+        sb.append('\n');
+
+        // Separator row
+        sb.append('|');
+        for (int c = 0; c < cols; c++) {
+            sb.append('-').append("-".repeat(widths[c])).append('-').append('|');
+        }
+        sb.append('\n');
+
+        // Data rows
+        for (int r = 1; r < grid.length; r++) {
+            sb.append(buildRow(grid[r], widths, cols));
+            if (r < grid.length - 1) {
+                sb.append('\n');
+            }
+        }
+
+        return sb.toString();
+    }
+
+    private static String buildRow(String[] row, int[] widths, int cols) {
+        StringBuilder sb = new StringBuilder();
+        sb.append('|');
+        for (int c = 0; c < cols; c++) {
+            String cell = c < row.length ? escape(row[c].trim()) : "";
+            sb.append(' ').append(padRight(cell, widths[c])).append(' ').append('|');
+        }
+        return sb.toString();
+    }
+
+    private static String escape(String cell) {
+        return cell.replace("|", "\\|");
+    }
+
+    private static String padRight(String s, int width) {
+        if (s.length() >= width) return s;
+        return s + " ".repeat(width - s.length());
+    }
+}

app/common/src/main/java/stirling/software/common/service/FileStorage.java

@@ -5,6 +5,7 @@ import java.io.IOException;
 import java.io.InputStream;
 import java.io.PipedInputStream;
 import java.io.PipedOutputStream;
+import java.util.Optional;
 import java.util.concurrent.Executors;
 import java.util.concurrent.atomic.AtomicReference;
 
@@ -17,6 +18,7 @@ import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 
 import stirling.software.common.cluster.FileStore;
+import stirling.software.common.util.JobContext;
 
 /**
  * Service for storing and retrieving files with unique file IDs. Used by the AutoJobPostMapping
@@ -32,8 +34,10 @@ public class FileStorage {
 
     private final FileOrUploadService fileOrUploadService;
     private final FileStore fileStore;
+    private final Optional<JobOwnershipService> jobOwnershipService;
 
     public String storeFile(MultipartFile file) throws IOException {
+        String owner = resolveOwner();
         // Fast path: when Spring buffered the multipart to disk (typical for large uploads), the
         // backing Resource exposes a real File. Hand the Path to the FileStore so it can do a
         // file-to-file copy (Linux sendfile, no copy through Java heap) rather than streaming
@@ -48,7 +52,7 @@ public class FileStorage {
         if (res != null && res.isFile()) {
             try {
                 FileStore.Stored stored =
-                        fileStore.store(res.getFile().toPath(), file.getOriginalFilename());
+                        fileStore.store(res.getFile().toPath(), file.getOriginalFilename(), owner);
                 log.debug("Stored file with ID: {} (fast path)", stored.fileId());
                 return stored.fileId();
             } catch (IOException ex) {
@@ -57,40 +61,45 @@ public class FileStorage {
             }
         }
         try (InputStream in = file.getInputStream()) {
-            FileStore.Stored stored = fileStore.store(in, file.getOriginalFilename());
+            FileStore.Stored stored = fileStore.store(in, file.getOriginalFilename(), owner);
             log.debug("Stored file with ID: {}", stored.fileId());
             return stored.fileId();
         }
     }
 
     public String storeBytes(byte[] bytes, String originalName) throws IOException {
-        FileStore.Stored stored = fileStore.store(new ByteArrayInputStream(bytes), originalName);
+        FileStore.Stored stored =
+                fileStore.store(new ByteArrayInputStream(bytes), originalName, resolveOwner());
         log.debug("Stored byte array with ID: {}", stored.fileId());
         return stored.fileId();
     }
 
     public MultipartFile retrieveFile(String fileId) throws IOException {
+        enforceOwnership(fileId);
         byte[] fileData = fileStore.retrieveBytes(fileId);
         return fileOrUploadService.toMockMultipartFile(fileId, fileData);
     }
 
     public byte[] retrieveBytes(String fileId) throws IOException {
+        enforceOwnership(fileId);
         return fileStore.retrieveBytes(fileId);
     }
 
     public InputStream retrieveInputStream(String fileId) throws IOException {
+        enforceOwnership(fileId);
         return fileStore.retrieve(fileId);
     }
 
     public StoredFile storeInputStream(InputStream inputStream, String originalName)
             throws IOException {
-        FileStore.Stored stored = fileStore.store(inputStream, originalName);
+        FileStore.Stored stored = fileStore.store(inputStream, originalName, resolveOwner());
         log.debug("Stored input stream with ID: {}", stored.fileId());
         return new StoredFile(stored.fileId(), stored.size());
     }
 
     public String storeFromStreamingBody(StreamingResponseBody body, String originalName)
             throws IOException {
+        String owner = resolveOwner();
         // Hold Throwable not IOException: an unchecked failure (NPE, IllegalState, OOM, etc.)
         // from the body writer would otherwise close the pipe with EOF and the consumer would
         // return a truncated file with no error surfaced to the caller.
@@ -115,7 +124,7 @@ public class FileStorage {
                                         }
                                     }
                                 });
-                FileStore.Stored stored = fileStore.store(in, originalName);
+                FileStore.Stored stored = fileStore.store(in, originalName, owner);
                 Throwable writerErr = bodyError.get();
                 if (writerErr != null) {
                     // Body failed mid-write: the FileStore persisted a truncated entry.
@@ -159,21 +168,62 @@ public class FileStorage {
 
     public String storeFromResource(Resource resource, String originalName) throws IOException {
         try (InputStream in = resource.getInputStream()) {
-            FileStore.Stored stored = fileStore.store(in, originalName);
+            FileStore.Stored stored = fileStore.store(in, originalName, resolveOwner());
             log.debug("Stored Resource with ID: {}", stored.fileId());
             return stored.fileId();
         }
     }
 
     public boolean deleteFile(String fileId) {
+        enforceOwnership(fileId);
         return fileStore.delete(fileId);
     }
 
     public boolean fileExists(String fileId) {
+        enforceOwnership(fileId);
         return fileStore.exists(fileId);
     }
 
     public long getFileSize(String fileId) throws IOException {
+        enforceOwnership(fileId);
         return fileStore.size(fileId);
     }
+
+    private String resolveOwner() {
+        String propagated = JobContext.getOwner();
+        if (propagated != null) {
+            return propagated;
+        }
+        return jobOwnershipService.flatMap(JobOwnershipService::getCurrentUserId).orElse(null);
+    }
+
+    private void enforceOwnership(String fileId) {
+        if (jobOwnershipService.isEmpty()) {
+            return;
+        }
+        Optional<String> currentUser = jobOwnershipService.get().getCurrentUserId();
+        if (currentUser.isEmpty()) {
+            return;
+        }
+        String owner;
+        try {
+            owner = fileStore.getOwner(fileId);
+        } catch (IOException e) {
+            log.warn("Failed to read owner for file {}: {}", fileId, e.getMessage());
+            throw new SecurityException(
+                    "Access denied: could not verify ownership of the requested file");
+        }
+        if (owner == null) {
+            return;
+        }
+        if (!owner.equals(currentUser.get())) {
+            log.warn(
+                    "Access denied: user {} attempted to access file {} owned by {}",
+                    currentUser.get(),
+                    fileId,
+                    owner);
+            throw new SecurityException(
+                    "Access denied: you do not have permission to access this file");
+        }
+    }
 }

app/common/src/main/java/stirling/software/common/service/JobExecutorService.java

@@ -89,6 +89,11 @@ public class JobExecutorService {
 
         String jobId = scopedJobKey;
 
+        final String jobOwner =
+                jobOwnershipService != null
+                        ? jobOwnershipService.getCurrentUserId().orElse(null)
+                        : null;
+
         long timeoutToUse = customTimeoutMs > 0 ? customTimeoutMs : effectiveTimeoutMs;
 
         log.debug(
@@ -119,6 +124,7 @@ public class JobExecutorService {
                         try {
                             stirling.software.common.util.JobContext.setJobId(
                                     capturedJobIdForQueue);
+                            stirling.software.common.util.JobContext.setOwner(jobOwner);
                             Object result = work.get();
                             processJobResult(capturedJobIdForQueue, result);
                             return result;
@@ -153,6 +159,7 @@ public class JobExecutorService {
                                     timeoutToUse);
 
                             stirling.software.common.util.JobContext.setJobId(capturedJobId);
+                            stirling.software.common.util.JobContext.setOwner(jobOwner);
                             Object result = executeWithTimeout(() -> work.get(), timeoutToUse);
                             processJobResult(capturedJobId, result);
                         } catch (TimeoutException te) {

app/common/src/main/java/stirling/software/common/service/ToolMetadataService.java

@@ -1,11 +1,21 @@
 package stirling.software.common.service;
 
+import java.util.List;
+
 /** Provides metadata about tool endpoints for internal dispatch. */
 public interface ToolMetadataService {
 
     /** Returns true if the given operation path accepts multiple input files. */
     boolean isMultiInput(String operationPath);
 
+    /**
+     * Returns the file extensions (lowercase, no leading dot, e.g. {@code "pdf"}) that the
+     * operation accepts as input ({@code output=false}) or produces as output ({@code
+     * output=true}), derived from the endpoint's declared type. Returns {@code null} when the
+     * endpoint declares no specific type, which callers should treat as "any type accepted".
+     */
+    List<String> getExtensionTypes(boolean output, String operationPath);
+
     /**
      * Returns true when the endpoint's ZIP response is a transport for multiple typed results and
      * should be unpacked: multi-output endpoints (Type:SIMO / Type:MIMO) and wrapper declarations

app/common/src/main/java/stirling/software/common/util/JobContext.java

@@ -1,8 +1,9 @@
 package stirling.software.common.util;
 
-/** Thread-local context for passing job ID across async boundaries */
+/** Thread-local context for passing job ID and owner across async boundaries */
 public class JobContext {
     private static final ThreadLocal<String> CURRENT_JOB_ID = new ThreadLocal<>();
+    private static final ThreadLocal<String> CURRENT_OWNER = new ThreadLocal<>();
 
     public static void setJobId(String jobId) {
         CURRENT_JOB_ID.set(jobId);
@@ -12,7 +13,16 @@ public class JobContext {
         return CURRENT_JOB_ID.get();
     }
 
+    public static void setOwner(String owner) {
+        CURRENT_OWNER.set(owner);
+    }
+
+    public static String getOwner() {
+        return CURRENT_OWNER.get();
+    }
+
     public static void clear() {
         CURRENT_JOB_ID.remove();
+        CURRENT_OWNER.remove();
     }
 }

app/common/src/main/java/stirling/software/common/util/OfficeDocumentSanitizer.java

@@ -0,0 +1,310 @@
+package stirling.software.common.util;
+
+import java.io.ByteArrayInputStream;
+import java.io.ByteArrayOutputStream;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Locale;
+import java.util.Set;
+import java.util.zip.ZipEntry;
+import java.util.zip.ZipInputStream;
+import java.util.zip.ZipOutputStream;
+
+import javax.xml.XMLConstants;
+import javax.xml.parsers.DocumentBuilder;
+import javax.xml.parsers.DocumentBuilderFactory;
+import javax.xml.parsers.ParserConfigurationException;
+import javax.xml.transform.OutputKeys;
+import javax.xml.transform.Transformer;
+import javax.xml.transform.TransformerException;
+import javax.xml.transform.TransformerFactory;
+import javax.xml.transform.dom.DOMSource;
+import javax.xml.transform.stream.StreamResult;
+
+import org.springframework.stereotype.Component;
+import org.w3c.dom.Document;
+import org.w3c.dom.Element;
+import org.w3c.dom.NamedNodeMap;
+import org.w3c.dom.Node;
+import org.w3c.dom.NodeList;
+import org.xml.sax.SAXException;
+
+import io.github.pixee.security.ZipSecurity;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.service.SsrfProtectionService;
+
+// Strips external refs from OOXML/ODF uploads so LibreOffice can't be made to fetch them.
+@Component
+@Slf4j
+public class OfficeDocumentSanitizer {
+
+    private static final Set<String> OOXML_EXTENSIONS =
+            Set.of(
+                    "docx", "docm", "dotx", "dotm", "xlsx", "xlsm", "xltx", "xltm", "pptx", "pptm",
+                    "potx", "potm", "ppsx", "ppsm");
+
+    private static final Set<String> ODF_EXTENSIONS =
+            Set.of(
+                    "odt", "ott", "ods", "ots", "odp", "otp", "odg", "otg", "odf", "odc", "odi",
+                    "odm");
+
+    private static final Set<String> ODF_XML_PARTS =
+            Set.of("content.xml", "styles.xml", "meta.xml", "settings.xml");
+
+    private final SsrfProtectionService ssrfProtectionService;
+    private final ApplicationProperties applicationProperties;
+
+    public OfficeDocumentSanitizer(
+            SsrfProtectionService ssrfProtectionService,
+            ApplicationProperties applicationProperties) {
+        this.ssrfProtectionService = ssrfProtectionService;
+        this.applicationProperties = applicationProperties;
+    }
+
+    public boolean isSanitizableExtension(String extension) {
+        if (extension == null) {
+            return false;
+        }
+        String lower = extension.toLowerCase(Locale.ROOT);
+        return OOXML_EXTENSIONS.contains(lower) || ODF_EXTENSIONS.contains(lower);
+    }
+
+    public byte[] sanitize(byte[] documentBytes, String extension) throws IOException {
+        if (documentBytes == null || documentBytes.length == 0) {
+            throw new IOException("Office document input is empty or null");
+        }
+        if (applicationProperties.getSystem().isDisableSanitize()) {
+            log.debug("Office document sanitization disabled by configuration");
+            return documentBytes;
+        }
+        if (!isSanitizableExtension(extension)) {
+            return documentBytes;
+        }
+
+        ByteArrayOutputStream out = new ByteArrayOutputStream(documentBytes.length);
+        try (ZipInputStream zipIn =
+                        ZipSecurity.createHardenedInputStream(
+                                new ByteArrayInputStream(documentBytes));
+                ZipOutputStream zipOut = new ZipOutputStream(out)) {
+
+            ZipEntry entry;
+            while ((entry = zipIn.getNextEntry()) != null) {
+                String name = entry.getName();
+                byte[] bytes = entry.isDirectory() ? new byte[0] : zipIn.readAllBytes();
+
+                if (!entry.isDirectory()) {
+                    bytes = sanitizeEntry(name, bytes);
+                }
+
+                ZipEntry outEntry = new ZipEntry(name);
+                if (entry.getComment() != null) {
+                    outEntry.setComment(entry.getComment());
+                }
+                if (entry.getExtra() != null) {
+                    outEntry.setExtra(entry.getExtra());
+                }
+                zipOut.putNextEntry(outEntry);
+                if (!entry.isDirectory()) {
+                    zipOut.write(bytes);
+                }
+                zipOut.closeEntry();
+            }
+        }
+        return out.toByteArray();
+    }
+
+    private byte[] sanitizeEntry(String entryName, byte[] entryBytes) {
+        String lower = entryName.toLowerCase(Locale.ROOT);
+        try {
+            if (lower.endsWith(".rels")) {
+                return sanitizeOoxmlRels(entryBytes);
+            }
+            if (isOdfXmlPart(lower)) {
+                return sanitizeOdfXml(entryBytes);
+            }
+        } catch (ParserConfigurationException
+                | SAXException
+                | IOException
+                | TransformerException e) {
+            log.warn(
+                    "Failed to parse XML part '{}' for sanitization, leaving as-is: {}",
+                    entryName,
+                    e.getMessage());
+        }
+        return entryBytes;
+    }
+
+    private boolean isOdfXmlPart(String lowerName) {
+        int slash = lowerName.lastIndexOf('/');
+        String base = slash >= 0 ? lowerName.substring(slash + 1) : lowerName;
+        return ODF_XML_PARTS.contains(base);
+    }
+
+    private byte[] sanitizeOoxmlRels(byte[] xmlBytes)
+            throws IOException, ParserConfigurationException, SAXException, TransformerException {
+        Document doc = parseSecurely(xmlBytes);
+        Element root = doc.getDocumentElement();
+        if (root == null) {
+            return xmlBytes;
+        }
+        NodeList relationships = root.getElementsByTagNameNS("*", "Relationship");
+        List<Node> toRemove = new ArrayList<>();
+        for (int i = 0; i < relationships.getLength(); i++) {
+            Node node = relationships.item(i);
+            NamedNodeMap attrs = node.getAttributes();
+            if (attrs == null) {
+                continue;
+            }
+            Node targetMode = attrs.getNamedItem("TargetMode");
+            if (targetMode == null || !"external".equalsIgnoreCase(targetMode.getNodeValue())) {
+                continue;
+            }
+            Node target = attrs.getNamedItem("Target");
+            String targetValue = target == null ? "" : target.getNodeValue();
+            if (isAdminAllowed(targetValue)) {
+                continue;
+            }
+            log.warn(
+                    "Stripping OOXML external relationship target: {}",
+                    truncateForLog(targetValue));
+            toRemove.add(node);
+        }
+        if (toRemove.isEmpty()) {
+            return xmlBytes;
+        }
+        for (Node n : toRemove) {
+            n.getParentNode().removeChild(n);
+        }
+        return serializeDocument(doc);
+    }
+
+    private byte[] sanitizeOdfXml(byte[] xmlBytes)
+            throws IOException, ParserConfigurationException, SAXException, TransformerException {
+        Document doc = parseSecurely(xmlBytes);
+        Element root = doc.getDocumentElement();
+        if (root == null) {
+            return xmlBytes;
+        }
+        boolean modified = stripExternalHrefs(root);
+        if (!modified) {
+            return xmlBytes;
+        }
+        return serializeDocument(doc);
+    }
+
+    private boolean stripExternalHrefs(Node node) {
+        boolean modified = false;
+        if (node.getNodeType() == Node.ELEMENT_NODE) {
+            NamedNodeMap attrs = node.getAttributes();
+            List<String> hrefAttrsToRemove = new ArrayList<>();
+            for (int i = 0; i < attrs.getLength(); i++) {
+                Node attr = attrs.item(i);
+                String name = attr.getNodeName();
+                if (name == null) {
+                    continue;
+                }
+                String lower = name.toLowerCase(Locale.ROOT);
+                if (!(lower.equals("xlink:href")
+                        || lower.endsWith(":href")
+                        || lower.equals("href"))) {
+                    continue;
+                }
+                String value = attr.getNodeValue();
+                if (!isExternalUrl(value)) {
+                    continue;
+                }
+                if (isAdminAllowed(value)) {
+                    continue;
+                }
+                log.warn(
+                        "Stripping ODF external href attribute ({}): {}",
+                        name,
+                        truncateForLog(value));
+                hrefAttrsToRemove.add(name);
+            }
+            Element element = (Element) node;
+            for (String attrName : hrefAttrsToRemove) {
+                element.removeAttribute(attrName);
+                modified = true;
+            }
+        }
+        NodeList children = node.getChildNodes();
+        for (int i = 0; i < children.getLength(); i++) {
+            if (stripExternalHrefs(children.item(i))) {
+                modified = true;
+            }
+        }
+        return modified;
+    }
+
+    private boolean isExternalUrl(String url) {
+        if (url == null) {
+            return false;
+        }
+        String trimmed = url.trim().toLowerCase(Locale.ROOT);
+        if (trimmed.isEmpty() || trimmed.startsWith("#") || trimmed.startsWith("../")) {
+            return false;
+        }
+        return trimmed.startsWith("http://")
+                || trimmed.startsWith("https://")
+                || trimmed.startsWith("ftp://")
+                || trimmed.startsWith("ftps://")
+                || trimmed.startsWith("file:")
+                || trimmed.startsWith("smb:")
+                || trimmed.startsWith("\\\\")
+                || trimmed.startsWith("//");
+    }
+
+    // Preserved only with an explicit allowedDomains entry; MEDIUM default would admit public URLs.
+    private boolean isAdminAllowed(String url) {
+        if (ssrfProtectionService == null || url == null || url.isBlank()) {
+            return false;
+        }
+        ApplicationProperties.Html.UrlSecurity config =
+                applicationProperties.getSystem().getHtml().getUrlSecurity();
+        if (config == null
+                || config.getAllowedDomains() == null
+                || config.getAllowedDomains().isEmpty()) {
+            return false;
+        }
+        return ssrfProtectionService.isUrlAllowed(url);
+    }
+
+    private Document parseSecurely(byte[] xmlBytes)
+            throws ParserConfigurationException, SAXException, IOException {
+        DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
+        factory.setFeature(XMLConstants.FEATURE_SECURE_PROCESSING, true);
+        factory.setFeature("http://apache.org/xml/features/disallow-doctype-decl", true);
+        factory.setFeature("http://xml.org/sax/features/external-general-entities", false);
+        factory.setFeature("http://xml.org/sax/features/external-parameter-entities", false);
+        factory.setFeature("http://apache.org/xml/features/nonvalidating/load-external-dtd", false);
+        factory.setXIncludeAware(false);
+        factory.setExpandEntityReferences(false);
+        factory.setNamespaceAware(true);
+        DocumentBuilder builder = factory.newDocumentBuilder();
+        return builder.parse(new ByteArrayInputStream(xmlBytes));
+    }
+
+    private byte[] serializeDocument(Document doc) throws TransformerException {
+        TransformerFactory tf = TransformerFactory.newInstance();
+        tf.setFeature(XMLConstants.FEATURE_SECURE_PROCESSING, true);
+        Transformer transformer = tf.newTransformer();
+        transformer.setOutputProperty(OutputKeys.ENCODING, "UTF-8");
+        transformer.setOutputProperty(OutputKeys.INDENT, "no");
+        transformer.setOutputProperty(OutputKeys.OMIT_XML_DECLARATION, "no");
+        ByteArrayOutputStream baos = new ByteArrayOutputStream();
+        transformer.transform(new DOMSource(doc), new StreamResult(baos));
+        return baos.toByteArray();
+    }
+
+    private String truncateForLog(String value) {
+        if (value == null) {
+            return "null";
+        }
+        return value.length() > 80 ? value.substring(0, 80) + "..." : value;
+    }
+}

app/common/src/test/java/stirling/software/SPDF/pdf/parser/LineAlignmentTableParserTest.java

@@ -1,153 +0,0 @@
-package stirling.software.SPDF.pdf.parser;
-
-import static org.assertj.core.api.Assertions.assertThat;
-import static stirling.software.SPDF.pdf.parser.PdfModels.*;
-
-import java.util.List;
-
-import org.junit.jupiter.api.Test;
-
-/**
- * Unit tests for {@link LineAlignmentTableParser}, focused on the coincident-line merge logic and
- * column-grid construction.
- */
-class LineAlignmentTableParserTest {
-
-    private final LineAlignmentTableParser parser = new LineAlignmentTableParser();
-
-    // ── mergeCoincidentLines ─────────────────────────────────────────────────────────────────────
-
-    @Test
-    void mergeCoincidentLines_singleLine_unchanged() {
-        var lines = List.of(tokenized(rawLine(10f, 100f, "Revenue")));
-        assertThat(parser.mergeCoincidentLines(lines)).hasSize(1);
-    }
-
-    @Test
-    void mergeCoincidentLines_distinctYLines_unchanged() {
-        // Two lines at different y positions — must NOT be merged.
-        var lines =
-                List.of(
-                        tokenized(rawLine(10f, 100f, "Revenue")),
-                        tokenized(rawLine(10f, 115f, "Cost")));
-        assertThat(parser.mergeCoincidentLines(lines)).hasSize(2);
-    }
-
-    @Test
-    void mergeCoincidentLines_sameY_merged() {
-        // Simulates a financial-table row split by LineBuilder at the column gap:
-        //   label fragment at x=72  → "Revenue"
-        //   value fragment at x=350 → "1,234"
-        // Both have y=100. After merge they should form one TokenizedLine.
-        var label = rawLine(72f, 100f, "Revenue");
-        var value = rawLine(350f, 100f, "1,234");
-
-        var merged = parser.mergeCoincidentLines(List.of(tokenized(label), tokenized(value)));
-
-        assertThat(merged).hasSize(1);
-        // The merged line should contain tokens from both halves.
-        var tokens = merged.get(0).all();
-        assertThat(tokens.stream().map(t -> t.text()).toList())
-                .containsExactlyInAnyOrder("Revenue", "1,234");
-    }
-
-    @Test
-    void mergeCoincidentLines_sameY_mergedLineHasCorrectBounds() {
-        var label = rawLine(72f, 100f, "Revenue"); // 7 chars × 6pt = 42pt wide → right = 114
-        var value = rawLine(350f, 100f, "1,234"); // 5 chars × 6pt = 30pt wide → right = 380
-
-        var merged = parser.mergeCoincidentLines(List.of(tokenized(label), tokenized(value)));
-
-        var bounds = merged.get(0).line().bounds();
-        assertThat(bounds.x()).isEqualTo(72f);
-        assertThat(bounds.right()).isEqualTo(380f);
-    }
-
-    @Test
-    void mergeCoincidentLines_withinTolerance_merged() {
-        // Lines 1.5pt apart (within ROW_MERGE_TOLERANCE_PT = 2pt) should merge.
-        var a = rawLine(10f, 100.0f, "Alpha");
-        var b = rawLine(200f, 101.5f, "99");
-
-        var merged = parser.mergeCoincidentLines(List.of(tokenized(a), tokenized(b)));
-        assertThat(merged).hasSize(1);
-    }
-
-    @Test
-    void mergeCoincidentLines_beyondTolerance_notMerged() {
-        // Lines 3pt apart (beyond ROW_MERGE_TOLERANCE_PT = 2pt) should NOT merge.
-        var a = rawLine(10f, 100.0f, "Alpha");
-        var b = rawLine(200f, 103.0f, "99");
-
-        var merged = parser.mergeCoincidentLines(List.of(tokenized(a), tokenized(b)));
-        assertThat(merged).hasSize(2);
-    }
-
-    @Test
-    void mergeCoincidentLines_threeCoincident_allMerged() {
-        // Three fragments at the same y (e.g. wide financial table with two value columns).
-        var a = rawLine(72f, 100f, "Revenue");
-        var b = rawLine(300f, 100f, "1,234");
-        var c = rawLine(400f, 100f, "5,678");
-
-        var merged = parser.mergeCoincidentLines(List.of(tokenized(a), tokenized(b), tokenized(c)));
-        assertThat(merged).hasSize(1);
-        assertThat(merged.get(0).all()).hasSize(3);
-    }
-
-    @Test
-    void mergeCoincidentLines_coincidentPairFollowedByDistinctLine_twoGroups() {
-        var a = rawLine(72f, 100f, "Revenue");
-        var b = rawLine(350f, 100f, "1,234"); // same y as a → merges with a
-        var c = rawLine(10f, 115f, "Expenses"); // different y → stays separate
-
-        var merged = parser.mergeCoincidentLines(List.of(tokenized(a), tokenized(b), tokenized(c)));
-        assertThat(merged).hasSize(2);
-    }
-
-    @Test
-    void mergeCoincidentLines_numericAnchorStatus_correctAfterMerge() {
-        // After merging, the combined line should be an anchor (≥2 numeric tokens).
-        // "Revenue" alone → not an anchor. "1,234  567" alone → anchor.
-        // Merged → anchor with at least 2 numerics.
-        var label = rawLine(72f, 100f, "Revenue");
-        var values = rawLineMultiWord(350f, 100f, "1,234", 30f, "567", 30f);
-
-        var merged = parser.mergeCoincidentLines(List.of(tokenized(label), tokenized(values)));
-
-        assertThat(merged).hasSize(1);
-        assertThat(merged.get(0).isAnchor()).isTrue();
-    }
-
-    // ── helpers ──────────────────────────────────────────────────────────────────────────────────
-
-    /** Creates a RawLine with a single TextFragment of the given text at the given position. */
-    private static RawLine rawLine(float x, float y, String text) {
-        float width = text.length() * 6f; // ~6pt per char — rough but consistent
-        float height = 12f;
-        Bounds bounds = new Bounds(x, y, width, height);
-        TextFragment fragment =
-                new TextFragment("tf-test", text, bounds, y + height, 11f, "Helvetica", false);
-        return new RawLine("ln-test", List.of(fragment), bounds, 1);
-    }
-
-    /**
-     * Creates a RawLine with two TextFragments representing two words separated by a small gap.
-     * Used to simulate a values-only line with multiple numeric tokens.
-     */
-    private static RawLine rawLineMultiWord(
-            float x, float y, String word1, float w1, String word2, float w2) {
-        float height = 12f;
-        Bounds b1 = new Bounds(x, y, w1, height);
-        Bounds b2 = new Bounds(x + w1 + 5f, y, w2, height);
-        TextFragment f1 = new TextFragment("tf-1", word1, b1, y + height, 11f, "Helvetica", false);
-        TextFragment f2 = new TextFragment("tf-2", word2, b2, y + height, 11f, "Helvetica", false);
-        Bounds lineBounds = new Bounds(x, y, x + w1 + 5f + w2 - x, height);
-        return new RawLine("ln-test", List.of(f1, f2), lineBounds, 1);
-    }
-
-    /** Tokenises a RawLine via the parser's own tokenise logic (package-private access). */
-    private LineAlignmentTableParser.TokenizedLine tokenized(RawLine line) {
-        return parser.tokenize(line);
-    }
-}

app/common/src/test/java/stirling/software/common/cluster/inprocess/LocalDiskFileStoreTest.java

@@ -3,11 +3,13 @@ package stirling.software.common.cluster.inprocess;
 import static org.junit.jupiter.api.Assertions.assertArrayEquals;
 import static org.junit.jupiter.api.Assertions.assertEquals;
 import static org.junit.jupiter.api.Assertions.assertFalse;
+import static org.junit.jupiter.api.Assertions.assertNull;
 import static org.junit.jupiter.api.Assertions.assertThrows;
 import static org.junit.jupiter.api.Assertions.assertTrue;
 
 import java.io.ByteArrayInputStream;
 import java.io.IOException;
+import java.nio.file.Files;
 import java.nio.file.Path;
 
 import org.junit.jupiter.api.Test;
@@ -39,4 +41,46 @@ class LocalDiskFileStoreTest {
         assertThrows(IllegalArgumentException.class, () -> store.resolve("a/b"));
         assertThrows(IllegalArgumentException.class, () -> store.resolve("a\\b"));
     }
+
+    @Test
+    void ownerSidecarCannotBeReadAsFileId(@TempDir Path dir) throws IOException {
+        LocalDiskFileStore store = new LocalDiskFileStore(dir.toString());
+        FileStore.Stored stored =
+                store.store(new ByteArrayInputStream("hi".getBytes()), "f.bin", "alice");
+        String sidecarId = stored.fileId() + ".owner";
+        assertThrows(IllegalArgumentException.class, () -> store.resolve(sidecarId));
+        assertThrows(IllegalArgumentException.class, () -> store.retrieveBytes(sidecarId));
+    }
+
+    @Test
+    void ownerIsPersistedAndReturnedByGetOwner(@TempDir Path dir) throws IOException {
+        LocalDiskFileStore store = new LocalDiskFileStore(dir.toString());
+        FileStore.Stored stored =
+                store.store(new ByteArrayInputStream("hi".getBytes()), "f.bin", "alice");
+        assertEquals("alice", store.getOwner(stored.fileId()));
+    }
+
+    @Test
+    void getOwnerReturnsNullWhenNoOwnerWasRecorded(@TempDir Path dir) throws IOException {
+        LocalDiskFileStore store = new LocalDiskFileStore(dir.toString());
+        FileStore.Stored stored =
+                store.store(new ByteArrayInputStream("hi".getBytes()), "f.bin", null);
+        assertNull(store.getOwner(stored.fileId()));
+    }
+
+    @Test
+    void getOwnerReturnsNullForUnknownFileId(@TempDir Path dir) throws IOException {
+        LocalDiskFileStore store = new LocalDiskFileStore(dir.toString());
+        assertNull(store.getOwner("00000000-0000-0000-0000-000000000000"));
+    }
+
+    @Test
+    void deleteRemovesOwnerSidecar(@TempDir Path dir) throws IOException {
+        LocalDiskFileStore store = new LocalDiskFileStore(dir.toString());
+        FileStore.Stored stored =
+                store.store(new ByteArrayInputStream("hi".getBytes()), "f.bin", "alice");
+        assertTrue(store.delete(stored.fileId()));
+        assertFalse(Files.exists(dir.resolve(stored.fileId() + ".owner")));
+        assertNull(store.getOwner(stored.fileId()));
+    }
 }

app/common/src/test/java/stirling/software/common/pdf/PdfMarkdownConverterTest.java

@@ -0,0 +1,269 @@
+package stirling.software.common.pdf;
+
+import static org.junit.jupiter.api.Assertions.assertDoesNotThrow;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+import static org.junit.jupiter.api.Assertions.fail;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.charset.StandardCharsets;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.stream.Stream;
+
+import org.junit.jupiter.api.Disabled;
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.io.TempDir;
+import org.junit.jupiter.params.ParameterizedTest;
+import org.junit.jupiter.params.provider.Arguments;
+import org.junit.jupiter.params.provider.MethodSource;
+
+import stirling.software.jpdfium.PdfDocument;
+import stirling.software.jpdfium.text.TextLine;
+import stirling.software.jpdfium.text.TextWord;
+
+/**
+ * Accuracy and robustness tests for {@link PdfMarkdownConverter}, comparing conversion output
+ * against hand-authored golden Markdown for a set of owned/synthetic fixtures.
+ *
+ * <p>The {@link #gatedFixtures()} set is enforced in CI: those fixtures currently convert within
+ * the accuracy threshold and guard against regressions. Fixtures still being iterated on live in
+ * {@link #wipFixtures()} under a {@link Disabled} test so the goldens stay in the tree without
+ * breaking the build. Enable the WIP test locally to see per-fixture scores while working on the
+ * converter.
+ */
+class PdfMarkdownConverterTest {
+
+    /** Accuracy threshold: output must share at least this fraction of content with the golden. */
+    private static final double THRESHOLD = 0.95;
+
+    @TempDir Path tmp;
+
+    /** Fixtures that meet the accuracy threshold today and therefore gate CI. */
+    static Stream<Arguments> gatedFixtures() {
+        return Stream.of(
+                Arguments.of("multi-column-test_lorem.pdf", "multi-column-test_lorem.md"),
+                Arguments.of("bordered-table-test_widget.pdf", "bordered-table-test_widget.md"),
+                Arguments.of("many-tables-test_stress.pdf", "many-tables-test_stress.md"));
+    }
+
+    /** Fixtures still below the threshold; tracked here, enable locally to iterate. */
+    static Stream<Arguments> wipFixtures() {
+        return Stream.of(
+                Arguments.of(
+                        "wrapped-cell-test_expense-report.pdf",
+                        "wrapped-cell-test_expense-report.md"));
+    }
+
+    @ParameterizedTest(name = "{0}")
+    @MethodSource("gatedFixtures")
+    void convertMatchesGoldenMarkdown(String pdfName, String mdName) throws IOException {
+        assertConversionMatchesGolden(pdfName, mdName);
+    }
+
+    @Disabled("WIP fixtures below the accuracy threshold; enable locally to iterate")
+    @ParameterizedTest(name = "{0}")
+    @MethodSource("wipFixtures")
+    void convertMatchesGoldenMarkdownWip(String pdfName, String mdName) throws IOException {
+        assertConversionMatchesGolden(pdfName, mdName);
+    }
+
+    /**
+     * Degenerate/extreme geometry must not crash the converter. A crafted or malformed PDF can
+     * position text anywhere via a text matrix, so a row's words can span from near the origin to a
+     * coordinate beyond {@link Integer#MAX_VALUE}. The old column-detection code sized an {@code
+     * int[]} straight from {@code (int) Math.ceil(maxX) - lo}, which either allocated a multi-GB
+     * array (OutOfMemoryError) or overflowed to a negative length (NegativeArraySizeException) —
+     * taking down the request thread. Detection must instead bail out and return no columns.
+     */
+    @Test
+    void columnDetectionSurvivesDegenerateGeometry() {
+        // x ≈ 2.5e9 is past Integer.MAX_VALUE; combined with a near-origin word it yields an
+        // implausible span that the pre-fix code turned into a fatal array allocation.
+        List<TextLine> rows = new ArrayList<>();
+        for (int r = 0; r < 4; r++) {
+            float y = 400f - r * 12f;
+            TextWord near = new TextWord(List.of(), 50f, y, 30f, 10f);
+            TextWord far = new TextWord(List.of(), 2_500_000_000f, y, 30f, 10f);
+            rows.add(new TextLine(List.of(near, far), 50f, y, 2_499_999_980f, 10f));
+        }
+
+        List<float[]> columns =
+                assertDoesNotThrow(() -> PdfMarkdownConverter.findColumnRangesFromLines(rows));
+        assertTrue(
+                columns.isEmpty(),
+                "implausible page span should disable column detection, not allocate from it");
+    }
+
+    private void assertConversionMatchesGolden(String pdfName, String mdName) throws IOException {
+        Path pdfPath = tmp.resolve(pdfName);
+        try (InputStream in =
+                getClass().getResourceAsStream("/pdf-ingestion-fixtures/" + pdfName)) {
+            if (in == null) {
+                fail("Fixture not found on classpath: /pdf-ingestion-fixtures/" + pdfName);
+            }
+            Files.copy(in, pdfPath);
+        }
+
+        String actual;
+        try (PdfDocument doc = PdfDocument.open(pdfPath)) {
+            actual = new PdfMarkdownConverter().convert(doc);
+        }
+
+        String expected;
+        try (InputStream in = getClass().getResourceAsStream("/pdf-ingestion-fixtures/" + mdName)) {
+            if (in == null) {
+                fail("Golden file not found on classpath: /pdf-ingestion-fixtures/" + mdName);
+            }
+            expected = new String(in.readAllBytes(), StandardCharsets.UTF_8);
+        }
+
+        // Image placeholders are not scored: their body text is a TODO ("ideally, add the info
+        // available about the image...") rather than real content, so comparing it would penalise
+        // output for matching a placeholder we intend to replace. Drop those lines from both sides.
+        expected = stripImagePlaceholders(expected);
+        actual = stripImagePlaceholders(actual);
+
+        double similarity = similarity(expected, actual);
+        if (similarity < THRESHOLD) {
+            fail(
+                    String.format(
+                            "Markdown output differs from golden file '%s' by %.1f%% (threshold %.0f%%):%n%s",
+                            mdName,
+                            (1.0 - similarity) * 100,
+                            (1.0 - THRESHOLD) * 100,
+                            unifiedDiff(expected, actual)));
+        }
+    }
+
+    /** Substring identifying an image-placeholder line, which is excluded from scoring. */
+    private static final String IMAGE_PLACEHOLDER_MARKER = "Image intentionally redacted";
+
+    /**
+     * Removes non-content lines from the comparison: image placeholders (TODO text we intend to
+     * replace) and GFM table separator rows (the {@code |---|---|} divider, whose exact dash count
+     * is cosmetic — any run of three or more dashes is valid Markdown).
+     */
+    private static String stripImagePlaceholders(String md) {
+        StringBuilder sb = new StringBuilder();
+        for (String line : md.split("\n", -1)) {
+            if (line.contains(IMAGE_PLACEHOLDER_MARKER)
+                    || line.strip().startsWith("<image redacted")
+                    || isTableSeparatorRow(line)) {
+                continue;
+            }
+            if (sb.length() > 0) {
+                sb.append('\n');
+            }
+            sb.append(line);
+        }
+        return sb.toString();
+    }
+
+    /** True for a GFM table separator row, e.g. {@code |---|:--:|---|} (only |, -, :, space). */
+    private static boolean isTableSeparatorRow(String line) {
+        String t = line.strip();
+        if (!t.contains("-")) {
+            return false;
+        }
+        return t.chars().allMatch(c -> c == '|' || c == '-' || c == ':' || c == ' ');
+    }
+
+    /**
+     * Character-level similarity: proportion of expected characters that appear in the LCS. O(n*m)
+     * but golden files are small enough that this is fine.
+     */
+    private static double similarity(String expected, String actual) {
+        if (expected.isEmpty() && actual.isEmpty()) return 1.0;
+        if (expected.isEmpty() || actual.isEmpty()) return 0.0;
+        // Strip all whitespace for a content-focused comparison
+        String e = expected.replaceAll("\\s+", " ").strip();
+        String a = actual.replaceAll("\\s+", " ").strip();
+        int lcs = lcsLength(e, a);
+        return (double) lcs / Math.max(e.length(), a.length());
+    }
+
+    private static int lcsLength(String a, String b) {
+        // Use two-row DP to keep memory reasonable
+        int m = a.length(), n = b.length();
+        int[] prev = new int[n + 1];
+        int[] curr = new int[n + 1];
+        for (int i = 1; i <= m; i++) {
+            for (int j = 1; j <= n; j++) {
+                if (a.charAt(i - 1) == b.charAt(j - 1)) {
+                    curr[j] = prev[j - 1] + 1;
+                } else {
+                    curr[j] = Math.max(curr[j - 1], prev[j]);
+                }
+            }
+            int[] tmp = prev;
+            prev = curr;
+            curr = tmp;
+            java.util.Arrays.fill(curr, 0);
+        }
+        return prev[n];
+    }
+
+    private static String unifiedDiff(String expected, String actual) {
+        String[] expectedLines = expected.split("\n", -1);
+        String[] actualLines = actual.split("\n", -1);
+
+        List<String> diff = new ArrayList<>();
+        diff.add("--- expected");
+        diff.add("+++ actual");
+
+        int maxLines = Math.max(expectedLines.length, actualLines.length);
+        int context = 3;
+        boolean inHunk = false;
+        int hunkStart = -1;
+        List<String> hunkLines = new ArrayList<>();
+
+        for (int i = 0; i < maxLines; i++) {
+            String exp = i < expectedLines.length ? expectedLines[i] : null;
+            String act = i < actualLines.length ? actualLines[i] : null;
+
+            boolean changed = exp == null || act == null || !exp.equals(act);
+            if (changed) {
+                if (!inHunk) {
+                    inHunk = true;
+                    hunkStart = Math.max(0, i - context);
+                    // add context lines before change
+                    for (int c = hunkStart; c < i; c++) {
+                        hunkLines.add(" " + (c < expectedLines.length ? expectedLines[c] : ""));
+                    }
+                }
+                if (exp != null) hunkLines.add("-" + exp);
+                if (act != null) hunkLines.add("+" + act);
+            } else {
+                if (inHunk) {
+                    hunkLines.add(" " + exp);
+                    // check if we're far enough past the last change to close the hunk
+                    boolean moreChanges = false;
+                    for (int j = i + 1; j < Math.min(i + context, maxLines); j++) {
+                        String e2 = j < expectedLines.length ? expectedLines[j] : null;
+                        String a2 = j < actualLines.length ? actualLines[j] : null;
+                        if (e2 == null || a2 == null || !e2.equals(a2)) {
+                            moreChanges = true;
+                            break;
+                        }
+                    }
+                    if (!moreChanges && (i - hunkStart) >= context) {
+                        diff.add("@@ -" + (hunkStart + 1) + " @@");
+                        diff.addAll(hunkLines);
+                        hunkLines.clear();
+                        inHunk = false;
+                    }
+                }
+            }
+        }
+
+        if (inHunk && !hunkLines.isEmpty()) {
+            diff.add("@@ -" + (hunkStart + 1) + " @@");
+            diff.addAll(hunkLines);
+        }
+
+        return String.join("\n", diff);
+    }
+}

app/common/src/test/java/stirling/software/common/service/FileStorageDelegationTest.java

@@ -5,6 +5,7 @@ import static org.mockito.Mockito.mock;
 
 import java.io.IOException;
 import java.nio.file.Path;
+import java.util.Optional;
 
 import org.junit.jupiter.api.Test;
 import org.junit.jupiter.api.io.TempDir;
@@ -19,7 +20,8 @@ class FileStorageDelegationTest {
         FileStorage fs =
                 new FileStorage(
                         mock(FileOrUploadService.class),
-                        new LocalDiskFileStore(tempDir.toString()));
+                        new LocalDiskFileStore(tempDir.toString()),
+                        Optional.empty());
         byte[] payload = "round-trip".getBytes();
         String id = fs.storeBytes(payload, "x.bin");
         assertArrayEquals(payload, fs.retrieveBytes(id));

app/common/src/test/java/stirling/software/common/service/FileStorageOwnershipTest.java

@@ -0,0 +1,107 @@
+package stirling.software.common.service;
+
+import static org.junit.jupiter.api.Assertions.assertArrayEquals;
+import static org.junit.jupiter.api.Assertions.assertThrows;
+import static org.mockito.Mockito.mock;
+import static org.mockito.Mockito.when;
+
+import java.io.IOException;
+import java.nio.file.Path;
+import java.util.Optional;
+import java.util.concurrent.atomic.AtomicReference;
+
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.io.TempDir;
+
+import stirling.software.common.cluster.inprocess.LocalDiskFileStore;
+import stirling.software.common.util.JobContext;
+
+class FileStorageOwnershipTest {
+
+    private FileStorage newStorageWithoutSecurity(Path tempDir) {
+        return new FileStorage(
+                mock(FileOrUploadService.class),
+                new LocalDiskFileStore(tempDir.toString()),
+                Optional.empty());
+    }
+
+    private FileStorage newStorageWithCurrentUser(Path tempDir, AtomicReference<String> userRef) {
+        JobOwnershipService svc = mock(JobOwnershipService.class);
+        when(svc.getCurrentUserId()).thenAnswer(invocation -> Optional.ofNullable(userRef.get()));
+        return new FileStorage(
+                mock(FileOrUploadService.class),
+                new LocalDiskFileStore(tempDir.toString()),
+                Optional.of(svc));
+    }
+
+    @Test
+    void desktopMode_noOwnershipService_storesAndRetrievesWithoutChecks(@TempDir Path tempDir)
+            throws IOException {
+        FileStorage fs = newStorageWithoutSecurity(tempDir);
+        byte[] payload = "desktop".getBytes();
+        String id = fs.storeBytes(payload, "x.bin");
+        assertArrayEquals(payload, fs.retrieveBytes(id));
+    }
+
+    @Test
+    void sameUserStoresAndRetrieves_allowed(@TempDir Path tempDir) throws IOException {
+        AtomicReference<String> user = new AtomicReference<>("alice");
+        FileStorage fs = newStorageWithCurrentUser(tempDir, user);
+        byte[] payload = "alice's file".getBytes();
+        String id = fs.storeBytes(payload, "x.bin");
+        assertArrayEquals(payload, fs.retrieveBytes(id));
+    }
+
+    @Test
+    void differentUserRetrieves_throwsSecurityException(@TempDir Path tempDir) throws IOException {
+        AtomicReference<String> user = new AtomicReference<>("alice");
+        FileStorage fs = newStorageWithCurrentUser(tempDir, user);
+        String id = fs.storeBytes("alice's file".getBytes(), "x.bin");
+        user.set("bob");
+        assertThrows(SecurityException.class, () -> fs.retrieveBytes(id));
+        assertThrows(SecurityException.class, () -> fs.retrieveInputStream(id));
+        assertThrows(SecurityException.class, () -> fs.getFileSize(id));
+        assertThrows(SecurityException.class, () -> fs.fileExists(id));
+        assertThrows(SecurityException.class, () -> fs.deleteFile(id));
+    }
+
+    @Test
+    void anonymousRetrieveOfOwnedFile_allowed_noCurrentUserMeansNoCompare(@TempDir Path tempDir)
+            throws IOException {
+        AtomicReference<String> user = new AtomicReference<>("alice");
+        FileStorage fs = newStorageWithCurrentUser(tempDir, user);
+        byte[] payload = "alice's file".getBytes();
+        String id = fs.storeBytes(payload, "x.bin");
+        user.set(null);
+        assertArrayEquals(payload, fs.retrieveBytes(id));
+    }
+
+    @Test
+    void authedRetrieveOfAnonymousFile_allowed_noOwnerOnFile(@TempDir Path tempDir)
+            throws IOException {
+        AtomicReference<String> user = new AtomicReference<>(null);
+        FileStorage fs = newStorageWithCurrentUser(tempDir, user);
+        byte[] payload = "no-owner".getBytes();
+        String id = fs.storeBytes(payload, "x.bin");
+        user.set("alice");
+        assertArrayEquals(payload, fs.retrieveBytes(id));
+    }
+
+    @Test
+    void propagatedOwner_scopesAsyncWriteWithNoLiveUser(@TempDir Path tempDir) throws IOException {
+        AtomicReference<String> user = new AtomicReference<>(null);
+        FileStorage fs = newStorageWithCurrentUser(tempDir, user);
+        byte[] payload = "alice's async result".getBytes();
+        String id;
+        try {
+            JobContext.setOwner("alice");
+            id = fs.storeBytes(payload, "x.bin");
+        } finally {
+            JobContext.clear();
+        }
+        user.set("alice");
+        assertArrayEquals(payload, fs.retrieveBytes(id));
+        user.set("bob");
+        assertThrows(SecurityException.class, () -> fs.retrieveBytes(id));
+    }
+}

app/common/src/test/java/stirling/software/common/service/FileStorageTest.java

@@ -9,6 +9,8 @@ import java.io.InputStream;
 import java.nio.charset.StandardCharsets;
 import java.nio.file.Files;
 import java.nio.file.Path;
+import java.util.Optional;
+import java.util.UUID;
 import java.util.stream.Stream;
 
 import org.junit.jupiter.api.BeforeEach;
@@ -37,7 +39,10 @@ class FileStorageTest {
     void setUp() throws IOException {
         MockitoAnnotations.openMocks(this);
         fileStorage =
-                new FileStorage(fileOrUploadService, new LocalDiskFileStore(tempDir.toString()));
+                new FileStorage(
+                        fileOrUploadService,
+                        new LocalDiskFileStore(tempDir.toString()),
+                        Optional.empty());
 
         // Create a mock MultipartFile
         mockFile = mock(MultipartFile.class);
@@ -79,7 +84,7 @@ class FileStorageTest {
     void testRetrieveFile() throws IOException {
         // Arrange
         byte[] fileContent = "Test PDF content".getBytes();
-        String fileId = "test-file-1";
+        String fileId = UUID.randomUUID().toString();
         Path filePath = tempDir.resolve(fileId);
         Files.write(filePath, fileContent);
 
@@ -99,7 +104,7 @@ class FileStorageTest {
     void testRetrieveBytes() throws IOException {
         // Arrange
         byte[] fileContent = "Test PDF content".getBytes();
-        String fileId = "test-file-2";
+        String fileId = UUID.randomUUID().toString();
         Path filePath = tempDir.resolve(fileId);
         Files.write(filePath, fileContent);
 
@@ -113,7 +118,7 @@ class FileStorageTest {
     @Test
     void testRetrieveFile_FileNotFound() {
         // Arrange
-        String nonExistentFileId = "non-existent-file";
+        String nonExistentFileId = UUID.randomUUID().toString();
 
         // Act & Assert
         assertThrows(IOException.class, () -> fileStorage.retrieveFile(nonExistentFileId));
@@ -122,7 +127,7 @@ class FileStorageTest {
     @Test
     void testRetrieveBytes_FileNotFound() {
         // Arrange
-        String nonExistentFileId = "non-existent-file";
+        String nonExistentFileId = UUID.randomUUID().toString();
 
         // Act & Assert
         assertThrows(IOException.class, () -> fileStorage.retrieveBytes(nonExistentFileId));
@@ -132,7 +137,7 @@ class FileStorageTest {
     void testDeleteFile() throws IOException {
         // Arrange
         byte[] fileContent = "Test PDF content".getBytes();
-        String fileId = "test-file-3";
+        String fileId = UUID.randomUUID().toString();
         Path filePath = tempDir.resolve(fileId);
         Files.write(filePath, fileContent);
 
@@ -147,7 +152,7 @@ class FileStorageTest {
     @Test
     void testDeleteFile_FileNotFound() {
         // Arrange
-        String nonExistentFileId = "non-existent-file";
+        String nonExistentFileId = UUID.randomUUID().toString();
 
         // Act
         boolean result = fileStorage.deleteFile(nonExistentFileId);
@@ -160,7 +165,7 @@ class FileStorageTest {
     void testFileExists() throws IOException {
         // Arrange
         byte[] fileContent = "Test PDF content".getBytes();
-        String fileId = "test-file-4";
+        String fileId = UUID.randomUUID().toString();
         Path filePath = tempDir.resolve(fileId);
         Files.write(filePath, fileContent);
 
@@ -174,7 +179,7 @@ class FileStorageTest {
     @Test
     void testFileExists_FileNotFound() {
         // Arrange
-        String nonExistentFileId = "non-existent-file";
+        String nonExistentFileId = UUID.randomUUID().toString();
 
         // Act
         boolean result = fileStorage.fileExists(nonExistentFileId);

app/common/src/test/java/stirling/software/common/util/OfficeDocumentSanitizerTest.java

@@ -0,0 +1,370 @@
+package stirling.software.common.util;
+
+import static org.junit.jupiter.api.Assertions.assertArrayEquals;
+import static org.junit.jupiter.api.Assertions.assertFalse;
+import static org.junit.jupiter.api.Assertions.assertThrows;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+import static org.mockito.ArgumentMatchers.eq;
+import static org.mockito.Mockito.lenient;
+import static org.mockito.Mockito.mock;
+
+import java.io.ByteArrayInputStream;
+import java.io.ByteArrayOutputStream;
+import java.io.IOException;
+import java.nio.charset.StandardCharsets;
+import java.util.HashMap;
+import java.util.LinkedHashMap;
+import java.util.Map;
+import java.util.zip.ZipEntry;
+import java.util.zip.ZipInputStream;
+import java.util.zip.ZipOutputStream;
+
+import org.junit.jupiter.api.BeforeEach;
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.service.SsrfProtectionService;
+
+class OfficeDocumentSanitizerTest {
+
+    private static final String EXTERNAL_URL = "https://webhook.site/ssrf-callback";
+    private static final String INTERNAL_TARGET = "media/image1.png";
+
+    private static final String DOCX_RELS =
+            "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                    + "<Relationships xmlns=\"http://schemas.openxmlformats.org/package/2006/relationships\">"
+                    + "<Relationship Id=\"rId1\" Type=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships/image\""
+                    + " Target=\""
+                    + EXTERNAL_URL
+                    + "\" TargetMode=\"External\"/>"
+                    + "<Relationship Id=\"rId2\" Type=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships/image\""
+                    + " Target=\""
+                    + INTERNAL_TARGET
+                    + "\"/>"
+                    + "</Relationships>";
+
+    private static final String DOCX_DOCUMENT =
+            "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                    + "<w:document xmlns:w=\"http://schemas.openxmlformats.org/wordprocessingml/2006/main\">"
+                    + "<w:body><w:p/></w:body></w:document>";
+
+    private static final String ODF_CONTENT_EXTERNAL =
+            "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                    + "<office:document-content"
+                    + " xmlns:office=\"urn:oasis:names:tc:opendocument:xmlns:office:1.0\""
+                    + " xmlns:draw=\"urn:oasis:names:tc:opendocument:xmlns:drawing:1.0\""
+                    + " xmlns:xlink=\"http://www.w3.org/1999/xlink\">"
+                    + "<office:body><office:text>"
+                    + "<draw:frame><draw:image xlink:href=\""
+                    + EXTERNAL_URL
+                    + "\" xlink:type=\"simple\"/></draw:frame>"
+                    + "<draw:frame><draw:image xlink:href=\"Pictures/image1.png\" xlink:type=\"simple\"/></draw:frame>"
+                    + "</office:text></office:body></office:document-content>";
+
+    private SsrfProtectionService ssrfProtectionService;
+    private ApplicationProperties applicationProperties;
+    private OfficeDocumentSanitizer sanitizer;
+
+    @BeforeEach
+    void setUp() {
+        applicationProperties = new ApplicationProperties();
+        ssrfProtectionService = mock(SsrfProtectionService.class);
+        sanitizer = new OfficeDocumentSanitizer(ssrfProtectionService, applicationProperties);
+    }
+
+    @Test
+    void isSanitizableExtension_recognizesOoxmlAndOdf() {
+        assertTrue(sanitizer.isSanitizableExtension("docx"));
+        assertTrue(sanitizer.isSanitizableExtension("DOCX"));
+        assertTrue(sanitizer.isSanitizableExtension("xlsx"));
+        assertTrue(sanitizer.isSanitizableExtension("pptx"));
+        assertTrue(sanitizer.isSanitizableExtension("odt"));
+        assertTrue(sanitizer.isSanitizableExtension("ods"));
+        assertTrue(sanitizer.isSanitizableExtension("odp"));
+        assertFalse(sanitizer.isSanitizableExtension("pdf"));
+        assertFalse(sanitizer.isSanitizableExtension("html"));
+        assertFalse(sanitizer.isSanitizableExtension(""));
+        assertFalse(sanitizer.isSanitizableExtension(null));
+    }
+
+    @Test
+    void sanitize_stripsOoxmlExternalRelationship() throws IOException {
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("word/_rels/document.xml.rels", DOCX_RELS.getBytes(StandardCharsets.UTF_8));
+        entries.put("word/document.xml", DOCX_DOCUMENT.getBytes(StandardCharsets.UTF_8));
+        byte[] docx = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(docx, "docx");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String rels =
+                new String(result.get("word/_rels/document.xml.rels"), StandardCharsets.UTF_8);
+        assertFalse(rels.contains(EXTERNAL_URL), "External URL should be stripped from .rels");
+        assertFalse(
+                rels.toLowerCase().contains("targetmode=\"external\""),
+                "TargetMode=External relationship should be removed");
+        assertTrue(rels.contains(INTERNAL_TARGET), "Internal image target should be preserved");
+        assertArrayEquals(
+                DOCX_DOCUMENT.getBytes(StandardCharsets.UTF_8),
+                result.get("word/document.xml"),
+                "Non-rels entries must be untouched");
+    }
+
+    @Test
+    void sanitize_pptxExternalImageRelStripped() throws IOException {
+        String pptxRels =
+                "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                        + "<Relationships xmlns=\"http://schemas.openxmlformats.org/package/2006/relationships\">"
+                        + "<Relationship Id=\"rId1\" Type=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships/image\""
+                        + " Target=\""
+                        + EXTERNAL_URL
+                        + "\" TargetMode=\"External\"/>"
+                        + "</Relationships>";
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("ppt/slides/_rels/slide1.xml.rels", pptxRels.getBytes(StandardCharsets.UTF_8));
+        byte[] pptx = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(pptx, "pptx");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String rels =
+                new String(result.get("ppt/slides/_rels/slide1.xml.rels"), StandardCharsets.UTF_8);
+        assertFalse(rels.contains(EXTERNAL_URL));
+    }
+
+    @Test
+    void sanitize_xlsxExternalImageRelStripped() throws IOException {
+        String xlsxRels =
+                "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                        + "<Relationships xmlns=\"http://schemas.openxmlformats.org/package/2006/relationships\">"
+                        + "<Relationship Id=\"rId1\" Type=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships/image\""
+                        + " Target=\""
+                        + EXTERNAL_URL
+                        + "\" TargetMode=\"External\"/>"
+                        + "</Relationships>";
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put(
+                "xl/drawings/_rels/drawing1.xml.rels", xlsxRels.getBytes(StandardCharsets.UTF_8));
+        byte[] xlsx = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(xlsx, "xlsx");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String rels =
+                new String(
+                        result.get("xl/drawings/_rels/drawing1.xml.rels"), StandardCharsets.UTF_8);
+        assertFalse(rels.contains(EXTERNAL_URL));
+    }
+
+    @Test
+    void sanitize_odtStripsExternalXlinkHrefButKeepsInternal() throws IOException {
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("content.xml", ODF_CONTENT_EXTERNAL.getBytes(StandardCharsets.UTF_8));
+        String manifestXml =
+                "<?xml version=\"1.0\"?><manifest:manifest"
+                        + " xmlns:manifest=\"urn:oasis:names:tc:opendocument:xmlns:manifest:1.0\"/>";
+        entries.put("META-INF/manifest.xml", manifestXml.getBytes(StandardCharsets.UTF_8));
+        byte[] odt = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(odt, "odt");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String content = new String(result.get("content.xml"), StandardCharsets.UTF_8);
+        assertFalse(content.contains(EXTERNAL_URL), "External xlink:href should be stripped");
+        assertTrue(content.contains("Pictures/image1.png"), "Internal href should be preserved");
+    }
+
+    @Test
+    void sanitize_odsStripsExternalXlinkHref() throws IOException {
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("content.xml", ODF_CONTENT_EXTERNAL.getBytes(StandardCharsets.UTF_8));
+        byte[] ods = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(ods, "ods");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String content = new String(result.get("content.xml"), StandardCharsets.UTF_8);
+        assertFalse(content.contains(EXTERNAL_URL));
+    }
+
+    @Test
+    void sanitize_odpStripsExternalXlinkHrefInStylesXml() throws IOException {
+        String stylesXml =
+                "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                        + "<office:document-styles"
+                        + " xmlns:office=\"urn:oasis:names:tc:opendocument:xmlns:office:1.0\""
+                        + " xmlns:draw=\"urn:oasis:names:tc:opendocument:xmlns:drawing:1.0\""
+                        + " xmlns:xlink=\"http://www.w3.org/1999/xlink\">"
+                        + "<draw:image xlink:href=\""
+                        + EXTERNAL_URL
+                        + "\"/></office:document-styles>";
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("styles.xml", stylesXml.getBytes(StandardCharsets.UTF_8));
+        byte[] odp = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(odp, "odp");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String content = new String(result.get("styles.xml"), StandardCharsets.UTF_8);
+        assertFalse(content.contains(EXTERNAL_URL));
+    }
+
+    @Test
+    void sanitize_disabledByConfigReturnsOriginal() throws IOException {
+        applicationProperties.getSystem().setDisableSanitize(true);
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("word/_rels/document.xml.rels", DOCX_RELS.getBytes(StandardCharsets.UTF_8));
+        byte[] docx = zip(entries);
+
+        byte[] result = sanitizer.sanitize(docx, "docx");
+        assertArrayEquals(docx, result);
+    }
+
+    @Test
+    void sanitize_unrecognizedExtensionReturnsOriginal() throws IOException {
+        byte[] original = "irrelevant".getBytes(StandardCharsets.UTF_8);
+        byte[] result = sanitizer.sanitize(original, "pdf");
+        assertArrayEquals(original, result);
+    }
+
+    @Test
+    void sanitize_emptyInputThrows() {
+        assertThrows(IOException.class, () -> sanitizer.sanitize(new byte[0], "docx"));
+    }
+
+    @Test
+    void sanitize_nullInputThrows() {
+        assertThrows(IOException.class, () -> sanitizer.sanitize(null, "docx"));
+    }
+
+    @Test
+    void sanitize_preservesEntryWithExternalRefWhenAdminAllowsDomain() throws IOException {
+        applicationProperties
+                .getSystem()
+                .getHtml()
+                .getUrlSecurity()
+                .getAllowedDomains()
+                .add("webhook.site");
+        lenient().when(ssrfProtectionService.isUrlAllowed(eq(EXTERNAL_URL))).thenReturn(true);
+
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("word/_rels/document.xml.rels", DOCX_RELS.getBytes(StandardCharsets.UTF_8));
+        byte[] docx = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(docx, "docx");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String rels =
+                new String(result.get("word/_rels/document.xml.rels"), StandardCharsets.UTF_8);
+        assertTrue(rels.contains(EXTERNAL_URL), "Allow-listed external URL should be preserved");
+    }
+
+    @Test
+    void sanitize_doesNotConsultSsrfServiceWhenAllowedDomainsEmpty() throws IOException {
+        // Even if mock would say allowed, we should not invoke it when there is no allow-list,
+        // because MEDIUM default would let public URLs through and re-introduce the vulnerability.
+        lenient().when(ssrfProtectionService.isUrlAllowed(eq(EXTERNAL_URL))).thenReturn(true);
+
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("word/_rels/document.xml.rels", DOCX_RELS.getBytes(StandardCharsets.UTF_8));
+        byte[] docx = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(docx, "docx");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String rels =
+                new String(result.get("word/_rels/document.xml.rels"), StandardCharsets.UTF_8);
+        assertFalse(rels.contains(EXTERNAL_URL));
+    }
+
+    @Test
+    void sanitize_handlesNonXmlEntriesSafely() throws IOException {
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        byte[] imageBytes = new byte[] {(byte) 0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a};
+        entries.put("word/media/image1.png", imageBytes);
+        entries.put("word/_rels/document.xml.rels", DOCX_RELS.getBytes(StandardCharsets.UTF_8));
+        byte[] docx = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(docx, "docx");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        assertArrayEquals(imageBytes, result.get("word/media/image1.png"));
+    }
+
+    @Test
+    void sanitize_internalLinksKeptWhenNoExternalPresent() throws IOException {
+        String internalOnlyRels =
+                "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                        + "<Relationships xmlns=\"http://schemas.openxmlformats.org/package/2006/relationships\">"
+                        + "<Relationship Id=\"rId1\" Type=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships/image\""
+                        + " Target=\"media/image1.png\"/>"
+                        + "</Relationships>";
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put(
+                "word/_rels/document.xml.rels", internalOnlyRels.getBytes(StandardCharsets.UTF_8));
+        byte[] docx = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(docx, "docx");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String rels =
+                new String(result.get("word/_rels/document.xml.rels"), StandardCharsets.UTF_8);
+        assertTrue(rels.contains("media/image1.png"));
+    }
+
+    @Test
+    void sanitize_corruptZipProducesSafeOutput() throws IOException {
+        byte[] garbage = "this is not a zip file".getBytes(StandardCharsets.UTF_8);
+        byte[] result = sanitizer.sanitize(garbage, "docx");
+        Map<String, byte[]> entries = unzip(result);
+        assertTrue(entries.isEmpty(), "Garbage input must not yield exploitable entries");
+    }
+
+    @Test
+    void sanitize_relativeOdfPathsArePreserved() throws IOException {
+        String content =
+                "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
+                        + "<office:document-content"
+                        + " xmlns:office=\"urn:oasis:names:tc:opendocument:xmlns:office:1.0\""
+                        + " xmlns:draw=\"urn:oasis:names:tc:opendocument:xmlns:drawing:1.0\""
+                        + " xmlns:xlink=\"http://www.w3.org/1999/xlink\">"
+                        + "<draw:image xlink:href=\"../Pictures/image1.png\"/>"
+                        + "<draw:image xlink:href=\"#anchor\"/>"
+                        + "</office:document-content>";
+        Map<String, byte[]> entries = new LinkedHashMap<>();
+        entries.put("content.xml", content.getBytes(StandardCharsets.UTF_8));
+        byte[] odt = zip(entries);
+
+        byte[] cleaned = sanitizer.sanitize(odt, "odt");
+
+        Map<String, byte[]> result = unzip(cleaned);
+        String out = new String(result.get("content.xml"), StandardCharsets.UTF_8);
+        assertTrue(out.contains("../Pictures/image1.png"));
+        assertTrue(out.contains("#anchor"));
+    }
+
+    private static byte[] zip(Map<String, byte[]> entries) throws IOException {
+        ByteArrayOutputStream baos = new ByteArrayOutputStream();
+        try (ZipOutputStream zos = new ZipOutputStream(baos)) {
+            for (Map.Entry<String, byte[]> e : entries.entrySet()) {
+                ZipEntry entry = new ZipEntry(e.getKey());
+                zos.putNextEntry(entry);
+                zos.write(e.getValue());
+                zos.closeEntry();
+            }
+        }
+        return baos.toByteArray();
+    }
+
+    private static Map<String, byte[]> unzip(byte[] data) throws IOException {
+        Map<String, byte[]> entries = new HashMap<>();
+        try (ZipInputStream zis = new ZipInputStream(new ByteArrayInputStream(data))) {
+            ZipEntry e;
+            while ((e = zis.getNextEntry()) != null) {
+                entries.put(e.getName(), zis.readAllBytes());
+                zis.closeEntry();
+            }
+        }
+        return entries;
+    }
+}

app/common/src/test/resources/pdf-ingestion-fixtures/bordered-table-test_widget.md

@@ -0,0 +1,10 @@
+# Widget Inventory Report
+
+This report lists current stock levels for each warehouse.
+
+| Region | Units | Status |
+|---|---|---|
+| North | 1200 | OK |
+| South | 950 | Low |
+| East | 1430 | OK |
+| West | 875 | Low |

app/common/src/test/resources/pdf-ingestion-fixtures/bordered-table-test_widget.pdf

@@ -0,0 +1,74 @@
+%PDF-1.4
+%“Œ‹ž ReportLab Generated PDF document (opensource)
+1 0 obj
+<<
+/F1 2 0 R /F2 3 0 R
+>>
+endobj
+2 0 obj
+<<
+/BaseFont /Helvetica /Encoding /WinAnsiEncoding /Name /F1 /Subtype /Type1 /Type /Font
+>>
+endobj
+3 0 obj
+<<
+/BaseFont /Helvetica-Bold /Encoding /WinAnsiEncoding /Name /F2 /Subtype /Type1 /Type /Font
+>>
+endobj
+4 0 obj
+<<
+/Contents 8 0 R /MediaBox [ 0 0 612 792 ] /Parent 7 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+5 0 obj
+<<
+/PageMode /UseNone /Pages 7 0 R /Type /Catalog
+>>
+endobj
+6 0 obj
+<<
+/Author (\(anonymous\)) /CreationDate (D:20260603003133+01'00') /Creator (\(unspecified\)) /Keywords () /ModDate (D:20260603003133+01'00') /Producer (ReportLab PDF Library - \(opensource\)) 
+  /Subject (\(unspecified\)) /Title (\(anonymous\)) /Trapped /False
+>>
+endobj
+7 0 obj
+<<
+/Count 1 /Kids [ 4 0 R ] /Type /Pages
+>>
+endobj
+8 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 500
+>>
+stream
+Gas1[9i&Y\%))C:pc(6t3;pQ8%0T@tD+-Nf,MP8j(><jDruNO^VsZ'm=ue7^Ia=;l(`h-+jUU3'Knh#YjH-MEIFj6r!oqf+FjMh;(rCq>R4tMCHL4NI*\1+UNiI'V9NC%VeJKn/YI0J];XQt&X83?=ihrg<*Mcn1n!1nWcDaQPe\P"9gnJuHl(jf]JQgZ[,&^uobI4QF',k"*^S)3c;)GMWC(T'=",ErnS#U=YCUN0&q4+*KmK1Zd*NI\GQDiZUG7;PTja8lulb"\PWWO#WcfI[ZB:6s*3g$be%?JH<b[c(?4J1!2SLDmbmkh8.c&CU)B?Xu,cp$<hP@=fLc>(n`oaEJ[XE'%QW=HE04M<,;ERm[MS=uYF=nN3jG'f@#?O48Ia,6Y-3m&tTWVq1?DeiBkp.Ug*;lVZX`Z=P.eklHhNV;!R_?QOuoeJ<0%7idG7GM8boU$^>N.N,2^;25]0Z8M<<]XMCct>noC'Qfb?`*[Mo+,F9#>t~>endstream
+endobj
+xref
+0 9
+0000000000 65535 f 
+0000000061 00000 n 
+0000000102 00000 n 
+0000000209 00000 n 
+0000000321 00000 n 
+0000000514 00000 n 
+0000000582 00000 n 
+0000000862 00000 n 
+0000000921 00000 n 
+trailer
+<<
+/ID 
+[<ee376a59c53e3f2af87024f65fd4222c><ee376a59c53e3f2af87024f65fd4222c>]
+% ReportLab generated PDF document -- digest (opensource)
+
+/Info 6 0 R
+/Root 5 0 R
+/Size 9
+>>
+startxref
+1511
+%%EOF

app/common/src/test/resources/pdf-ingestion-fixtures/many-tables-test_stress.md

@@ -0,0 +1,222 @@
+Intro paragraph for section 1.
+
+| Name | Qty |
+|---|---|
+| alpha | 101 |
+| delta | 201 |
+
+# Section 2 Heading
+
+| Name | Qty | Price |
+|---|---|---|
+| alpha | 101 | charlie |
+| delta | 201 | foxtrot |
+| golf | 301 | india |
+
+## Section 3 Heading
+
+| Name | Qty | Price | Region |
+|---|---|---|---|
+| alpha | 101 | charlie | 3 |
+| delta | 201 | foxtrot | 13 |
+| golf | 301 | india | 23 |
+| juliet | 401 | lima | 33 |
+
+Intro paragraph for section 4.
+
+| Name | Qty | Price | Region | Status |
+|---|---|---|---|---|
+| alpha | 101 | charlie | 3 | echo |
+| delta | 201 | foxtrot | 13 | hotel |
+| golf | 301 | india | 23 | kilo |
+| juliet | 401 | lima | 33 | november |
+| mike | 501 | oscar | 43 | alpha |
+
+# Section 5 Heading
+
+| Name | Qty |
+|---|---|
+| alpha | 101 |
+| delta | 201 |
+| golf | 301 |
+| juliet | 401 |
+| mike | 501 |
+| papa | 601 |
+
+| Name | Qty | Price |
+|---|---|---|
+| alpha | 101 | charlie |
+| delta | 201 | foxtrot |
+
+# Section 7 Heading
+
+Intro paragraph for section 7.
+
+| Name | Qty | Price | Region |
+|---|---|---|---|
+| alpha | 101 | charlie | 3 |
+| delta | 201 | foxtrot | 13 |
+| golf | 301 | india | 23 |
+
+## Section 8 Heading
+
+| Name | Qty | Price | Region | Status |
+|---|---|---|---|---|
+| alpha | 101 | charlie | 3 | echo |
+| delta | 201 | foxtrot | 13 | hotel |
+| golf | 301 | india | 23 | kilo |
+| juliet | 401 | lima | 33 | november |
+
+| Name | Qty |
+|---|---|
+| alpha | 101 |
+| delta | 201 |
+| golf | 301 |
+| juliet | 401 |
+| mike | 501 |
+
+# Section 10 Heading
+
+Intro paragraph for section 10.
+
+| Name | Qty | Price |
+|---|---|---|
+| alpha | 101 | charlie |
+| delta | 201 | foxtrot |
+| golf | 301 | india |
+| juliet | 401 | lima |
+| mike | 501 | oscar |
+| papa | 601 | bravo |
+
+| Name | Qty | Price | Region |
+|---|---|---|---|
+| alpha | 101 | charlie | 3 |
+| delta | 201 | foxtrot | 13 |
+
+# Section 12 Heading
+
+| Name | Qty | Price | Region | Status |
+|---|---|---|---|---|
+| alpha | 101 | charlie | 3 | echo |
+| delta | 201 | foxtrot | 13 | hotel |
+| golf | 301 | india | 23 | kilo |
+
+## Section 13 Heading
+
+Intro paragraph for section 13.
+
+| Name | Qty |
+|---|---|
+| alpha | 101 |
+| delta | 201 |
+| golf | 301 |
+| juliet | 401 |
+
+| Name | Qty | Price |
+|---|---|---|
+| alpha | 101 | charlie |
+| delta | 201 | foxtrot |
+| golf | 301 | india |
+| juliet | 401 | lima |
+| mike | 501 | oscar |
+
+# Section 15 Heading
+
+| Name | Qty | Price | Region |
+|---|---|---|---|
+| alpha | 101 | charlie | 3 |
+| delta | 201 | foxtrot | 13 |
+| golf | 301 | india | 23 |
+| juliet | 401 | lima | 33 |
+| mike | 501 | oscar | 43 |
+| papa | 601 | bravo | 53 |
+
+Intro paragraph for section 16.
+
+| Name | Qty | Price | Region | Status |
+|---|---|---|---|---|
+| alpha | 101 | charlie | 3 | echo |
+| delta | 201 | foxtrot | 13 | hotel |
+
+# Section 17 Heading
+
+| Name | Qty |
+|---|---|
+| alpha | 101 |
+| delta | 201 |
+| golf | 301 |
+
+## Section 18 Heading
+
+| Name | Qty | Price |
+|---|---|---|
+| alpha | 101 | charlie |
+| delta | 201 | foxtrot |
+| golf | 301 | india |
+| juliet | 401 | lima |
+
+Intro paragraph for section 19.
+
+| Name | Qty | Price | Region |
+|---|---|---|---|
+| alpha | 101 | charlie | 3 |
+| delta | 201 | foxtrot | 13 |
+| golf | 301 | india | 23 |
+| juliet | 401 | lima | 33 |
+| mike | 501 | oscar | 43 |
+
+# Section 20 Heading
+
+| Name | Qty | Price | Region | Status |
+|---|---|---|---|---|
+| alpha | 101 | charlie | 3 | echo |
+| delta | 201 | foxtrot | 13 | hotel |
+| golf | 301 | india | 23 | kilo |
+| juliet | 401 | lima | 33 | november |
+| mike | 501 | oscar | 43 | alpha |
+| papa | 601 | bravo | 53 | delta |
+
+| Name | Qty |
+|---|---|
+| alpha | 101 |
+| delta | 201 |
+
+# Section 22 Heading
+
+Intro paragraph for section 22.
+
+| Name | Qty | Price |
+|---|---|---|
+| alpha | 101 | charlie |
+| delta | 201 | foxtrot |
+| golf | 301 | india |
+
+## Section 23 Heading
+
+| Name | Qty | Price | Region |
+|---|---|---|---|
+| alpha | 101 | charlie | 3 |
+| delta | 201 | foxtrot | 13 |
+| golf | 301 | india | 23 |
+| juliet | 401 | lima | 33 |
+
+| Name | Qty | Price | Region | Status |
+|---|---|---|---|---|
+| alpha | 101 | charlie | 3 | echo |
+| delta | 201 | foxtrot | 13 | hotel |
+| golf | 301 | india | 23 | kilo |
+| juliet | 401 | lima | 33 | november |
+| mike | 501 | oscar | 43 | alpha |
+
+# Section 25 Heading
+
+Intro paragraph for section 25.
+
+| Name | Qty |
+|---|---|
+| alpha | 101 |
+| delta | 201 |
+| golf | 301 |
+| juliet | 401 |
+| mike | 501 |
+| papa | 601 |

app/common/src/test/resources/pdf-ingestion-fixtures/many-tables-test_stress.pdf

@@ -0,0 +1,169 @@
+%PDF-1.4
+%“Œ‹ž ReportLab Generated PDF document (opensource)
+1 0 obj
+<<
+/F1 2 0 R /F2 3 0 R
+>>
+endobj
+2 0 obj
+<<
+/BaseFont /Helvetica /Encoding /WinAnsiEncoding /Name /F1 /Subtype /Type1 /Type /Font
+>>
+endobj
+3 0 obj
+<<
+/BaseFont /Helvetica-Bold /Encoding /WinAnsiEncoding /Name /F2 /Subtype /Type1 /Type /Font
+>>
+endobj
+4 0 obj
+<<
+/Contents 13 0 R /MediaBox [ 0 0 612 792 ] /Parent 12 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+5 0 obj
+<<
+/Contents 14 0 R /MediaBox [ 0 0 612 792 ] /Parent 12 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+6 0 obj
+<<
+/Contents 15 0 R /MediaBox [ 0 0 612 792 ] /Parent 12 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+7 0 obj
+<<
+/Contents 16 0 R /MediaBox [ 0 0 612 792 ] /Parent 12 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+8 0 obj
+<<
+/Contents 17 0 R /MediaBox [ 0 0 612 792 ] /Parent 12 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+9 0 obj
+<<
+/Contents 18 0 R /MediaBox [ 0 0 612 792 ] /Parent 12 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+10 0 obj
+<<
+/PageMode /UseNone /Pages 12 0 R /Type /Catalog
+>>
+endobj
+11 0 obj
+<<
+/Author (\(anonymous\)) /CreationDate (D:20260603005358+01'00') /Creator (\(unspecified\)) /Keywords () /ModDate (D:20260603005358+01'00') /Producer (ReportLab PDF Library - \(opensource\)) 
+  /Subject (\(unspecified\)) /Title (\(anonymous\)) /Trapped /False
+>>
+endobj
+12 0 obj
+<<
+/Count 6 /Kids [ 4 0 R 5 0 R 6 0 R 7 0 R 8 0 R 9 0 R ] /Type /Pages
+>>
+endobj
+13 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 1007
+>>
+stream
+GauHKgN)$k&:O:Sn<io8&IUqd[51N1e*/]GIR%m`>DZmeMUK]*(BdhV"2[X,j0;*\%_E*0HXJRc`<Z.PlPcH-)*5ON#Qr4-#h+U(ll5egXa[FW7^<l-+Mko1pU57W'%??d'1Q-,]\qWS80;JOp0HFGa1T[+D>q%4S#/VM`O[Rh4i5T.Phi,$]aT$05!q:PBuCQU/.p2]@1koN*Tb*K9k5G\ht+Dr\K=+8\NZ"alMaOEo**@OK:8-.O1X3-?Gg`@m3%,ti3'">T-&c=M&Wuu?cDbGDp.gO<KniopM)A\M"ajn9N]6p@tefHlX67:iW3SE/[O\LV1TKe.i9es,%kWY+1-s,ft.u27]:^>F0!r6&&CLC$tM?fIR"M37["/*k9@YkpKKSS@Np"1/4#R]`I^(g*1Sc,9L3Qt6N(T@F`A>oBGgL-!r#Y4)R\G`D,iVtFeJUE9u4iuUQ?D%C&SB4kkp.>D5tii>nDKJ"Y07jANhOb$R(_$=U7Stjs)-/KZ(6IBm`6<Z:IO1/g<Q6n^@fdK*]SO#g&Wj4B)g1,#O?30",57V_ONl0p`%uQN]'EAZV$475Z"'DApH2)Sg6UQ75XC?4^'<&6&d?YC`OF9F3eE;WIMt@a?'q)c\3oWQ_BD]d`4=O\,'aSUFV4HsE7)O:lPOi6Ht1,[dTm6JqQM6es[?9G4VC(YieF*)uM6[SZT#[TN2,BIir^1;/Ku"E1laH[\"M+F'8=n0!;?_#hHd!hkpV0u\Ij\4VC-km.[F&JhGmbc`6@I1=mf>u<3;i.Bh4+MFJ"H:.XWUQX6%LU(sg4Tt$_":5`p.2,gZkpUfdg,Hd(qR1)9\ltF2^8b5,,XbY%VfhD52O7A.c.u]dhTc5t%0<0L5E38`Bq+i;"%J7kcc#@i)@okdN-qiaA"33DdPgPrUs7;j%+47_cnGN@&ug9/dqGOAHA4f.N*&guHspf?E;GIc-Wt>:<(m1AmcS_Zc2VlEI'_S_>@#!MF^m1$$<mB1`gt\X~>endstream
+endobj
+14 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 987
+>>
+stream
+GatU3gN&c;&:O:SkV;H,eI!J\f?X"LSPIZ'"4Y.F#oAAY?N.Y?8Iu:s)eV;,aNG1Lea>G$!MSc\rU6m7jF'AOI09V1,^TSTi$A+f4sZVQ%2^>uj$40\AI>WbZo9q!NL^CS9P3E9X.:XEMb$a/X"qH9hN6fY,]D!OEOVT,722%RJnVqK4f'[4d3(/hbJWJRs,>>28jk5A:nCl/%FlC#LntsAC5cE$q`QO_@Pk%Lp>7V23")NNrFkcXfuoI=SJJ-`g)_+K\@,SQ@8ORGg(_(B.[u[WXD%Q8@I";9d(]M62K_)Q+-<\kHY5,)o99%BB2bcQVkd:+MZVH=$Z`S@BhH]-XfPANk@qBN.:Jc'?.Kn\p7(aPI<Cb@g:XRZ9Km+RD-e6S\N*kQMA?P5>QOI(du7U]WDrNNPW%d"8P_j^Y8d%G<t1\CJ_5-m+>,'JJLXrV.UD!ah19&9b#*fgR2o730<9)QX)X0"EK6u;mBJs3/fl\d&K$B3bXI5R>F*E,^\%+b?0o8s$o.CuX4uDAT_?G0(p4N3La,8qfi'j?e`e88KrZ8*NDgc:62'icCb],4B]Z"SO&e/BRC*C$2PmbSKAjc\$FLD8?&qHVH3G4p834/77n[%()-p/JDRBapcO2b\,3dW%#U9u!ilSkd%2'rr&g[u".1O]W%0J<@#(ptUD;%0:^.^ls:.#.V6NgR[`2\RUW!k$-*GZIf3DO<K8np=hCA>BiB_Mu"=LVGlO\1Z]0pF(@UaiVQ=dh?h5kfQ_gj>$jHW$8TCt:C.jVU98ne.XoiMtoV;q8XhnlOE;3a]=dGJ8KL&+Khj"&P4Q=P-:nHsDprY"B4+#A3dF,<W=NC>;b&gf(sT\Ts*iH)f!KhfHV;\mMSHej.9.XrBG41_<:b)MG;-Y(:&nCKS0F]r&CGl>EfMc0td\0GHScl3Lr;?OUoo6619Qdr)S8S);]7L5rmWW&(5t4Dil"USDhhY/Y=!=!X:85**$:~>endstream
+endobj
+15 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 976
+>>
+stream
+GatU3gN&c;'R]XVkgAc"70atdYFXp#3h7VV#7/=-%N$"G?N.YO&iX#cVsS_l_5i8)eum:1-)(7`mXJ7LnhqY0hGZal,We=q^e"$M]Lu;_=4AoIN$QOjOF.@i*Sg!_rLL=-'?kk;m;M&R`$D>'HCL7<qMHOgP.\NgCm:Zf5tLAgarh6YCKLONS%?!`3T6F7YKNJ(dBFh(U1"e7D6YdC:71/`?cjR"QrKtsGg*:'=Ck>]$A,-S`H1I/1T]4\B3=gQ9:I4j9f]+af*Z7*GH5H+;j73Z-T#DAcgpO<SFuA['l7@MF*6c^<pB]o84rG$&I@Cs3`KIG,N/bG0f<k/M@gUKb"%\W-lZ#AU+i=gUKkm46K-b>W(/'b?XF9+X+'-Dqi"+1U7?6=ZRJNBSJ6ojDTa+]o4RF^26C+i0s(9G2o;$@RNbs\(f75],L(SVpI0M!'7JVhZ"$t/%kb1+)F\p54/@jpVJF2+)7^F=gVcp8(h*m%F#'p9c%9Ep9?b7g^F5>(>I3t^d_"f'aue%^E\X4kN:g\P[;sC=g3h,nD/n9*`94uoM?sBCL<kr$AOqh:,Y&c90,npSDXN*bEfY(k^0LZ_\I@Y+L2nn64-qsuX5D'=o'B2"8rCsRJ?h/-4Ur6khb;ph%X"8-mXTUU5-!Uh$iVuRi1.=>4;XilfMPT4]&FC.U%DI(c6uf3*XPJP2MRdq@Ag3uYF().nqLFAE2Nih'(^p\5FnSNa>n+cI?U65A61_N1<9ZXH4Xr)Fgq(6md3H9[MQ3Q^%aq>E?XM:O:RKGjc+QV2MB!.I.17QpSt]X[C*6ap.`#'o;8F(ebZ.VMM6k`BkZB\I,2MNoX]J"k]Qd";,@($aWX&29q"6k;>Let)@]QHm:i/lf[DB_&U@ROq-0MLQp;@j+]?$"E)brXVun52AVQo$"[a7@!?!LT*>$&:5X_.R;9)'$l-S.>WjLQIVTo.JU%(H\,-*pl_kP9~>endstream
+endobj
+16 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 1024
+>>
+stream
+GauHKflEQ9'Rf^WgrHc4<#\/Qm7c-rFIIq+TFSD%Z#L'6o(Nl&^_!k0ds=.%-s&pt#i0QBKE91*<^/MJJCcfoHA_e;aL?\&_BAj]Dt;HW$6(8Ql\%O7-5%uT8<qrc9!'@hZ!'`C3a&iu\ZDZi+5lF*E@>0ZIL#\l)(+[ABOTNKmq!2F,S:.F93Z6QIu[nf\Q[qCE`Y-=.8k$67iCR!7=El+50a@f&b^6'Yg'mqlcRQ-5j:%[K@'5!A_m0L)&VlW1T50^X)"3Ma3,U1P8Di$>uPZ)Zj_igBc0lm]WE0#e>*M53(Yh6\9jHYh6B+3bM\b[Q9j'O_9:!:W0<K\bX;'ZV[%iHrL=:rH.Wh5O-II8-Rh3..Sd^\dNItUUd<nli^`$=A@?\WW4/7Wnp>$Ijm,#(bKn_D?UYMGJ7LLqC0\[$l<8U*O5^GnN;!N6YB'?d[d8DAPt^>+E:r0<2$'VL/TtuNg;C@5iM#%KdS-GU->0]`/20eLXVW#qnjsU.P:Xj&=enVE[mpqDCn2!qDRuQhI/#cZ-#m6`U9'SI4"9\<WSWoVPaf@]<NRl?FrTPOd:V_=>"r:4p.\W>IK_$#C,EkU#uO1g6*HoUK':XiJjtQfI'Sdc)B3J1eqgGJ+l?"$<J+^?VHgiq^90uD"@6ar+/_WhKG+8t!Slo>//&C$^f'TA!2:K!0pd;^Zif)c[:f!C/8hNKc7+'+1WbOFKr9T7c_G$s*O634O2Zs:"iN>j,pWU+:kEQ>5YbF=$27*59iJE<glrr=EAMssR&N"lVT1:*a,OcI=8\"6/>A//7(KsL\pUG.G[+JB-r8t;Sih+(W.A'8Q*LZK*I9WjCLjEt>lT%2RFJZg0Ps:5c,HdK#tSNI:#Rp!]O$Ic=T%K0[._E_Fcjn>OcM,iSAf9iF7,,+Qe:@o'Y4o_:54t*l+TAg<i_F?ke9:\G<n<b*t@"_`_;08ft@[9;qtKC#H0Z6cJ+d)MdidY-OFV1K"=!=K=^?hK-Q.s+?\[I98%IXRiAI.!:AVTf`~>endstream
+endobj
+17 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 1075
+>>
+stream
+GauHKgN&c;&:O:SkV;H,6&Q<FCM0)BVUK5trr=4p-3<-<P.#jI+b[BoQ`]$,GaorC,Bfp>Z[5)3-pL]0bpl':D91fR,o"F49.1/bfmFt32ljt6eO[j7!K]\nZ:^QQASu[l@3no(6*HL'@3I!Qfi7%,.N\.>A91O)D\n5*o@JWSOI@ng5p/"_I`g*Y$j0+5PWaso'25ltcglFd,QIaO'dPO0Yr\h5/,=m\PiP:P+Gg0&N;_5iLN$]BD.#VL+h,tSe\fL&,;+2fg&%\pAJiV#eVhq@ro%%/`[(&8n29YiYd>bDF^Bg[AVf/n>[FE(au"ZAe&>%Q[7/n3-QhsPXuPbp#<$d"UR8u-nBnr!Q$Wjd1;9WN/,KG8])Ca5rrDrI'Ue'*I0C%pn+f`Hd"4sB_&2*)S$pX=F/6"DC"TU`bMtd/m7u]?]=J5L[SNBuE:6Ri,NmW7.*qUp)97Tt+a84b\os`Ti8/uRD<mg2;n&q)23L#\f:[QJk+sYYhIC_O\TNS/;3<qfN@GE^L#jl^i&)"7_/%Cb;-M;;ooT&`R;C&\WTZ4"\YIcTZDa;0CKTX+?@ra\94$H:Q<l)r*NHFJWjMr7[tCBb^PX,G]H!fs0g3,5^'AZ%<4Zqp(JCd+;^lkoDJ%q(Bb84V=:PCgOtM#Rr-=oHW&DmO?/!mYrV;g]MR?14cGuH_H*]73!34(()[A.)c<V3:@I>lhrl*Xrn(a0\,%4%*k^rL\k3%6ako`&A3MoN^CXV77D!Qg<S_OEMIh5(T*nFt'&4]_38'Q<s*V7!;A%5-@W.E/(oCu;?^7kb[#R/-P*GF@KHDm6E*dLMg/4Q$W(<Ta[fVpGB5td>%=j9Ge+0B@dZDHELb2fDU(,iiXXRl^*U,U#Fld!R72UIpK<KU`cn&R/_eYS//1PE/$d7p:t-BP62fL<3/QQ4%=NoD?Tlf:A,GpJc<]*uYaL6fSAqm"-E.#%h]idd9$6\DZKKp8Vd[>GWu#-DXi653a?U$fqCs18gt(5/#U%!u,gkMK;>OT_/='S:kn,iD@u]<8-rlTRu'+\7L$U7-#!5-5\WQn(n:q@-p24u!~>endstream
+endobj
+18 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 590
+>>
+stream
+GasbXgMWc?&;KY!MRgt!"n`^Bf\9$c*`Q-Sb6tfd8P(q-dT1eng=QV-!OG*\kiWo2AGBdKLqI^%h,XNB)-gDk+GFVBa?g*a%:!J\97Vc8-jH8>8P>0S@Wr)o*"Hu<6qPp`EF:M3'l4@S\c2]`&[J%dO@5p$<(([H6:SlB;FS8L`&pO%6Y\V"/O7F]E%L@d.%>L@"4bK_XXs&";n_.W^Nr847HH,X@$.pC"4bYC?9RH,dN?h8<BTH9l89:QMq:es\ePUpFRc4?hV'h3b4`sE_b4r->a+8/!?Q=D\F41Nc@'B`';4XgMeh!;U23C+>AbMQD-o--lJ/":m#(Yt,X5(`1KL,GTMBLlr6=-1mi#k.-Ou\\T4$Pun2dEU(\?$&;1@T4dm^t!KuOD-U@N_AMr"&uVpsGm,+8I7B*f!%9.o4cC1<Nr^$d5!dH-#UIr,n6Z;JUFYWCTVA\,or>a[CZ12(hd>1*0bU`k2-MXo1[Gor#kmXGIM'#R49X#NSAOpdf'0ilUH4:M(^Snc3;m/+><UlJ1jSN&ZuoQ68F8`c/Y-e0[!aljuioC5fR7-VUAX`%g<0P5J7"CFlQJT1~>endstream
+endobj
+xref
+0 19
+0000000000 65535 f 
+0000000061 00000 n 
+0000000102 00000 n 
+0000000209 00000 n 
+0000000321 00000 n 
+0000000516 00000 n 
+0000000711 00000 n 
+0000000906 00000 n 
+0000001101 00000 n 
+0000001296 00000 n 
+0000001491 00000 n 
+0000001561 00000 n 
+0000001842 00000 n 
+0000001932 00000 n 
+0000003031 00000 n 
+0000004109 00000 n 
+0000005176 00000 n 
+0000006292 00000 n 
+0000007459 00000 n 
+trailer
+<<
+/ID 
+[<ba877f458edd6b06b69a7e843c67586d><ba877f458edd6b06b69a7e843c67586d>]
+% ReportLab generated PDF document -- digest (opensource)
+
+/Info 11 0 R
+/Root 10 0 R
+/Size 19
+>>
+startxref
+8140
+%%EOF

app/common/src/test/resources/pdf-ingestion-fixtures/multi-column-test_lorem.md

@@ -0,0 +1,25 @@
+# Lorem Ipsum in Two Columns
+
+## 1. Origins
+
+Lorem ipsum dolor sit amet consectetur adipiscing elit. Sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
+
+## 2. Structure
+
+Ut enim ad minim veniam quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
+
+## 3. Usage
+
+Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
+
+## 4. Variations
+
+Excepteur sint occaecat cupidatat non proident sunt in culpa qui officia deserunt mollit anim id est laborum.
+
+## 5. Typography
+
+Curabitur pretium tincidunt lacus. Nulla gravida orci a odio. Nullam various turpis et commodo pharetra est.
+
+## 6. Conclusion
+
+Nunc nonummy metus. Vestibulum volutpat pretium libero. Cras id dui. Aenean ut eros et nisl sagittis vestibulum.

app/common/src/test/resources/pdf-ingestion-fixtures/multi-column-test_lorem.pdf

@@ -0,0 +1,74 @@
+%PDF-1.4
+%“Œ‹ž ReportLab Generated PDF document (opensource)
+1 0 obj
+<<
+/F1 2 0 R /F2 3 0 R
+>>
+endobj
+2 0 obj
+<<
+/BaseFont /Helvetica /Encoding /WinAnsiEncoding /Name /F1 /Subtype /Type1 /Type /Font
+>>
+endobj
+3 0 obj
+<<
+/BaseFont /Helvetica-Bold /Encoding /WinAnsiEncoding /Name /F2 /Subtype /Type1 /Type /Font
+>>
+endobj
+4 0 obj
+<<
+/Contents 8 0 R /MediaBox [ 0 0 612 792 ] /Parent 7 0 R /Resources <<
+/Font 1 0 R /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ]
+>> /Rotate 0 /Trans <<
+
+>> 
+  /Type /Page
+>>
+endobj
+5 0 obj
+<<
+/PageMode /UseNone /Pages 7 0 R /Type /Catalog
+>>
+endobj
+6 0 obj
+<<
+/Author (\(anonymous\)) /CreationDate (D:20260603021636+01'00') /Creator (\(unspecified\)) /Keywords () /ModDate (D:20260603021636+01'00') /Producer (ReportLab PDF Library - \(opensource\)) 
+  /Subject (\(unspecified\)) /Title (\(anonymous\)) /Trapped /False
+>>
+endobj
+7 0 obj
+<<
+/Count 1 /Kids [ 4 0 R ] /Type /Pages
+>>
+endobj
+8 0 obj
+<<
+/Filter [ /ASCII85Decode /FlateDecode ] /Length 861
+>>
+stream
+Gat=i>Ak00&;B$5/*8Q/Z)fmNp(^2^OK'MC[dT6#Oq.&b^4c4;1No6>+D%TC.n+fice+l9R0[L%'(o<DJBpB+r$jsRcu23$RjI].E7rS*BB"3t\qRKaS-SYt[_<-]),Cp@'$i;6dQuJ#Q*+lGY?X<phgEWPa?op]_1g@kf_>sQ?dQP@AUa.A:T=#%ZIIl18hB7Zf<piVi-V+;]ct^jS<=Z4l:IOd[,o"%q.hK[[k]r1!^oX'%44os7T9[m%YrbZ3KU/;aR0$cV$;D._".OT8)jeEBKU'l<A`'aD&R'L"Bo]Q)R+=cKOEld8s#M0ZYZN4[lRm1*F6^d9-Gfl5>I(hVU+?`]UK%;aabAH:q>9NUGQq^.=u])Wt-ETI))C'a[FE@elj!RSrf2Q'F>URAO.C,!DneTPqrj#4e2kb9%1"4qfZ)"#&^0j9nHQ?9nF!j7mVPP5\*Uq'_jMVS]9%`kQB\8*AF_bpr/hGj;HCUOSQU-%5:6S79Ud\b!*tPbr_'pCr$Ea#(FYP31NFhSX.-("1M:$cgH#hX8L(2]R3Q>'BYHCS%pI!;=WdJp,'ii[`QPZ_9mcd\baZ2U(_;c\-p,8EoIEpQ*lstL>]LE;C#\dLnT2R:)BM-fTc['3_He[U,k'!Bo".uERd>SkhRj^J+koSIrZ_dEf_5L'/1h.`+DTK(R:P,WH)h5\se=SZ"L/5b8b..,e/E\o+4YQ+*im^C>AERG/TieEK\)#>U@HXnJ,H0A9-MqhhkDp8%.6Lr,OrK*lih;B<-opZ8%EU?,$r^jmCDAQ`-0/-8/`[p]7Fm%0f:E&S*FV)DX2>#q\bRqA=^_`43#8EA$u%8r6F5`rc9>K@q>E4q^*~>endstream
+endobj
+xref
+0 9
+0000000000 65535 f 
+0000000061 00000 n 
+0000000102 00000 n 
+0000000209 00000 n 
+0000000321 00000 n 
+0000000514 00000 n 
+0000000582 00000 n 
+0000000862 00000 n 
+0000000921 00000 n 
+trailer
+<<
+/ID 
+[<21a9fbd0a0991a91b6e6e2db0856056e><21a9fbd0a0991a91b6e6e2db0856056e>]
+% ReportLab generated PDF document -- digest (opensource)
+
+/Info 6 0 R
+/Root 5 0 R
+/Size 9
+>>
+startxref
+1872
+%%EOF

app/common/src/test/resources/pdf-ingestion-fixtures/wrapped-cell-test_expense-report.md

@@ -0,0 +1,62 @@
+# Employee Expense Report
+
+Reimbursement Request
+
+EMP-1047
+
+**Report Header**
+
+| Employee Name | Michael Tran |
+|---|---|
+| Employee ID | EMP-1047 |
+| Department | Client Services |
+| Report Date | January 20th, 2026 |
+| Reporting Period | January 5th–16th, 2026 |
+| Manager Approver | Laura Simmons |
+
+**Company Information**
+
+| Company | Summit Consulting Partners |
+|---|---|
+| Company Address | 88 Riverside Plaza, Suite 1400, New York, NY 10069 |
+| Accounting Department Email | expenses@example.com |
+
+**Trip Purpose**
+
+The trip was undertaken for client onsite meetings with Atlantic Energy Solutions in Boston, MA.
+
+**Expense Details**
+
+| Description | Amount | Date | Category |
+|---|---|---|---|
+| Flight (NYC to Boston roundtrip) | $325.40 | January 5th, 2026 | Airline ticket |
+| Hotel (3 nights at Harborview Hotel) | $822.75 | January 5th–8th, 2026 | Lodging |
+| Taxi from airport to hotel | $48.00 | January 5th, 2026 | Ground transportation |
+| Client dinner (3 attendees) | $186.20 | January 6th, 2026 | Meals |
+| Parking at JFK Airport | $72.00 | January 5th–8th, 2026 | Parking |
+| Breakfast (per diem not used) | $18.50 | January 7th, 2026 | Meals |
+
+| Description | Amount | Date | Category |
+|---|---|---|---|
+| Uber to client office | $22.10 | January 7th, 2026 | Ground transportation |
+| Printing + presentation materials | $46.90 | January 8th, 2026 | Materials |
+| Lunch with client | $39.75 | January 8th, 2026 | Meals |
+| Office supplies (notebooks, pens) | $27.60 | January 10th, 2026 | Supplies |
+| Mileage reimbursement (client visit in NJ, 42 miles @ $0.67/mile) | $28.14 | January 14th, 2026 | Mileage |
+| Team lunch meeting (internal) | $64.30 | January 15th, 2026 | Meals |
+
+Total Expenses $1,701.64
+
+Reimbursement Method
+
+Reimbursement method Direct deposit
+
+Notes
+
+All receipts are attached. Expenses are business-related and comply with company travel policy.
+
+**Approval**
+
+Michael Tran, Employee
+
+Laura Simmons, Manager

app/core/src/main/java/stirling/software/SPDF/config/OpenApiConfig.java

@@ -18,6 +18,7 @@ import io.swagger.v3.oas.models.media.StringSchema;
 import io.swagger.v3.oas.models.security.SecurityRequirement;
 import io.swagger.v3.oas.models.security.SecurityScheme;
 import io.swagger.v3.oas.models.servers.Server;
+import io.swagger.v3.oas.models.tags.Tag;
 
 import lombok.RequiredArgsConstructor;
 
@@ -60,6 +61,15 @@ public class OpenApiConfig {
 
         OpenAPI openAPI = new OpenAPI().info(info).openapi("3.0.3");
 
+        // Register a single global "AI" tag so every AI endpoint groups under it in the docs.
+        // The AI controllers are currently @Hidden, so they don't emit this tag themselves yet;
+        // defining it here keeps the grouping ready for when those endpoints are unhidden.
+        openAPI.addTagsItem(
+                new Tag()
+                        .name("AI")
+                        .description(
+                                "AI-powered document creation, editing, and assistant endpoints."));
+
         // Add server configuration from environment variable
         String swaggerServerUrl = System.getenv("SWAGGER_SERVER_URL");
         Server server;

app/core/src/main/java/stirling/software/SPDF/controller/api/PdfOverlayController.java

@@ -10,6 +10,7 @@ import java.util.Map;
 
 import org.apache.pdfbox.Loader;
 import org.apache.pdfbox.multipdf.Overlay;
+import org.apache.pdfbox.pdfwriter.compress.CompressParameters;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.springframework.core.io.Resource;
 import org.springframework.http.MediaType;
@@ -157,7 +158,10 @@ public class PdfOverlayController {
                     PDDocument singlePageDocument = new PDDocument()) {
                 singlePageDocument.addPage(overlayPdf.getPage(pageCountInCurrentOverlay));
                 File tempFile = Files.createTempFile("overlay-page-", ".pdf").toFile();
-                singlePageDocument.save(tempFile);
+                // NO_COMPRESSION: this single-page doc holds a page copied from overlayPdf.
+                // PDFBox 3.0.7's compressed writer (PDFBOX-6203) drops shared resources imported
+                // across documents, corrupting overlay fonts. Revert once on 3.0.8.
+                singlePageDocument.save(tempFile, CompressParameters.NO_COMPRESSION);
 
                 overlayGuide.put(basePageIndex, tempFile.getAbsolutePath());
                 tempFiles.add(tempFile); // Keep track of the temporary file for cleanup

app/core/src/main/java/stirling/software/SPDF/controller/api/RearrangePagesPDFController.java

@@ -6,11 +6,9 @@ import java.util.Collections;
 import java.util.List;
 import java.util.Locale;
 
-import org.apache.pdfbox.cos.COSName;
 import org.apache.pdfbox.pdmodel.PDDocument;
-import org.apache.pdfbox.pdmodel.PDDocumentCatalog;
 import org.apache.pdfbox.pdmodel.PDPage;
-import org.apache.pdfbox.pdmodel.interactive.form.PDAcroForm;
+import org.apache.pdfbox.pdmodel.PDPageTree;
 import org.springframework.core.io.Resource;
 import org.springframework.http.MediaType;
 import org.springframework.http.ResponseEntity;
@@ -262,38 +260,31 @@ public class RearrangePagesPDFController {
                 }
                 log.info("newPageOrder = {}", newPageOrder);
                 log.info("totalPages = {}", totalPages);
-                // Create a new list to hold the pages in the new order
-                List<PDPage> newPages = new ArrayList<>();
-                for (int i = 0; i < newPageOrder.size(); i++) {
-                    newPages.add(document.getPage(newPageOrder.get(i)));
+
+                // Snapshot the desired pages before mutating the source document's page tree.
+                List<PDPage> newPages = new ArrayList<>(newPageOrder.size());
+                for (Integer idx : newPageOrder) {
+                    newPages.add(document.getPage(idx));
                 }
 
-                // Create a new document based on the original one
-                try (PDDocument rearrangedDocument =
-                        pdfDocumentFactory.createNewDocumentBasedOnOldDocument(document)) {
-
-                    // Add the pages in the new order
-                    for (PDPage page : newPages) {
-                        rearrangedDocument.addPage(page);
-                    }
-
-                    PDDocumentCatalog sourceCatalog = document.getDocumentCatalog();
-                    if (sourceCatalog != null) {
-                        PDAcroForm sourceForm = sourceCatalog.getAcroForm(null);
-                        if (sourceForm != null) {
-                            rearrangedDocument
-                                    .getDocumentCatalog()
-                                    .getCOSObject()
-                                    .setItem(COSName.ACRO_FORM, sourceForm.getCOSObject());
-                        }
-                    }
-
-                    return WebResponseUtils.pdfDocToWebResponse(
-                            rearrangedDocument,
-                            GeneralUtils.generateFilename(
-                                    pdfFile.getOriginalFilename(), "_rearranged.pdf"),
-                            tempFileManager);
+                // Rearrange in-place on the source document rather than copying pages into a
+                // freshly-created PDDocument. Copying pages across documents triggers a PDFBox
+                // 3.0.7 compressed-save regression (PDFBOX-6203, fixed for 3.0.8) where shared
+                // resource objects (fonts, etc.) imported from the source can be silently
+                // dropped from the output, producing pages with "font not found" errors.
+                PDPageTree pages = document.getPages();
+                for (int i = totalPages - 1; i >= 0; i--) {
+                    pages.remove(i);
                 }
+                for (PDPage page : newPages) {
+                    pages.add(page);
+                }
+
+                return WebResponseUtils.pdfDocToWebResponse(
+                        document,
+                        GeneralUtils.generateFilename(
+                                pdfFile.getOriginalFilename(), "_rearranged.pdf"),
+                        tempFileManager);
             }
         } catch (IOException e) {
             ExceptionUtils.logException("document rearrangement", e);

app/core/src/main/java/stirling/software/SPDF/controller/api/converters/ConvertOfficeController.java

@@ -35,6 +35,7 @@ import stirling.software.common.service.CustomPDFDocumentFactory;
 import stirling.software.common.util.CustomHtmlSanitizer;
 import stirling.software.common.util.ExceptionUtils;
 import stirling.software.common.util.GeneralUtils;
+import stirling.software.common.util.OfficeDocumentSanitizer;
 import stirling.software.common.util.ProcessExecutor;
 import stirling.software.common.util.ProcessExecutor.ProcessExecutorResult;
 import stirling.software.common.util.RegexPatternUtils;
@@ -50,6 +51,7 @@ public class ConvertOfficeController {
     private final CustomPDFDocumentFactory pdfDocumentFactory;
     private final RuntimePathConfig runtimePathConfig;
     private final CustomHtmlSanitizer customHtmlSanitizer;
+    private final OfficeDocumentSanitizer officeDocumentSanitizer;
     private final EndpointConfiguration endpointConfiguration;
     private final TempFileManager tempFileManager;
 
@@ -83,14 +85,16 @@ public class ConvertOfficeController {
         Path inputPath = workDir.resolve(baseName + "." + extensionLower);
         Path outputPath = workDir.resolve(baseName + ".pdf");
 
-        // Check if the file is HTML and apply sanitization if needed
+        // Sanitize input before LibreOffice sees it so embedded URLs can't trigger SSRF.
         if ("html".equals(extensionLower) || "htm".equals(extensionLower)) {
-            // Read and sanitize HTML content
             String htmlContent = new String(inputFile.getBytes(), StandardCharsets.UTF_8);
             String sanitizedHtml = customHtmlSanitizer.sanitize(htmlContent);
             Files.writeString(inputPath, sanitizedHtml, StandardCharsets.UTF_8);
+        } else if (officeDocumentSanitizer.isSanitizableExtension(extensionLower)) {
+            byte[] sanitized =
+                    officeDocumentSanitizer.sanitize(inputFile.getBytes(), extensionLower);
+            Files.write(inputPath, sanitized);
         } else {
-            // copy file content
             Files.copy(inputFile.getInputStream(), inputPath, StandardCopyOption.REPLACE_EXISTING);
         }
 

app/core/src/main/java/stirling/software/SPDF/controller/api/misc/AutoSplitPdfController.java

@@ -14,6 +14,7 @@ import java.util.zip.ZipEntry;
 import java.util.zip.ZipOutputStream;
 
 import org.apache.pdfbox.cos.COSName;
+import org.apache.pdfbox.pdfwriter.compress.CompressParameters;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
 import org.apache.pdfbox.pdmodel.graphics.image.PDImageXObject;
@@ -357,7 +358,10 @@ public class AutoSplitPdfController {
                 for (int i = 0; i < splitDocuments.size(); i++) {
                     String fileName = filename + "_" + (i + 1) + ".pdf";
                     zipOut.putNextEntry(new ZipEntry(fileName));
-                    splitDocuments.get(i).save(zipOut);
+                    // NO_COMPRESSION: split docs are built by addPage()-ing pages copied from the
+                    // source document. PDFBox 3.0.7's compressed writer (PDFBOX-6203) drops shared
+                    // resources imported across documents, corrupting fonts. Revert once on 3.0.8.
+                    splitDocuments.get(i).save(zipOut, CompressParameters.NO_COMPRESSION);
                     zipOut.closeEntry();
                 }
             }

app/core/src/main/java/stirling/software/SPDF/controller/api/misc/OCRController.java

@@ -17,6 +17,7 @@ import javax.imageio.ImageIO;
 
 import org.apache.pdfbox.io.IOUtils;
 import org.apache.pdfbox.multipdf.PDFMergerUtility;
+import org.apache.pdfbox.pdfwriter.compress.CompressParameters;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
 import org.apache.pdfbox.rendering.PDFRenderer;
@@ -427,7 +428,10 @@ public class OCRController {
                             // Save original page without OCR as fallback
                             try (PDDocument pageDoc = new PDDocument()) {
                                 pageDoc.addPage(page);
-                                pageDoc.save(pageOutputPath);
+                                // NO_COMPRESSION: page is copied from another document;
+                                // PDFBox 3.0.7 compressed writer (PDFBOX-6203) drops shared
+                                // resources, corrupting fonts. Revert once on 3.0.8.
+                                pageDoc.save(pageOutputPath, CompressParameters.NO_COMPRESSION);
                             }
                         }
 
@@ -437,7 +441,10 @@ public class OCRController {
                         // Save original page without OCR
                         try (PDDocument pageDoc = new PDDocument()) {
                             pageDoc.addPage(page);
-                            pageDoc.save(pageOutputPath);
+                            // NO_COMPRESSION: page is copied from another document; PDFBox 3.0.7
+                            // compressed writer (PDFBOX-6203) drops shared resources, corrupting
+                            // fonts on retained text pages. Revert once on 3.0.8.
+                            pageDoc.save(pageOutputPath, CompressParameters.NO_COMPRESSION);
                             merger.addSource(pageOutputPath);
                         }
                     }

app/core/src/main/java/stirling/software/SPDF/controller/api/misc/OverlayImageController.java

@@ -25,6 +25,7 @@ import stirling.software.common.annotations.api.MiscApi;
 import stirling.software.common.enumeration.ResourceWeight;
 import stirling.software.common.service.CustomPDFDocumentFactory;
 import stirling.software.common.util.GeneralUtils;
+import stirling.software.common.util.SvgSanitizer;
 import stirling.software.common.util.TempFile;
 import stirling.software.common.util.TempFileManager;
 import stirling.software.common.util.WebResponseUtils;
@@ -36,6 +37,7 @@ public class OverlayImageController {
 
     private final CustomPDFDocumentFactory pdfDocumentFactory;
     private final TempFileManager tempFileManager;
+    private final SvgSanitizer svgSanitizer;
 
     @AutoJobPostMapping(
             consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
@@ -61,6 +63,9 @@ public class OverlayImageController {
             byte[] imageBytes = imageFile.getBytes();
 
             boolean isSvg = SvgOverlayUtil.isSvgImage(imageBytes);
+            if (isSvg) {
+                imageBytes = svgSanitizer.sanitize(imageBytes);
+            }
 
             try (PDDocument document = pdfDocumentFactory.load(pdfBytes)) {
                 int pages = document.getNumberOfPages();

app/core/src/main/java/stirling/software/SPDF/controller/web/ReactRoutingController.java

@@ -41,6 +41,8 @@ public class ReactRoutingController {
     private boolean indexHtmlExists = false;
     private boolean useExternalIndexHtml = false;
     private boolean loggedMissingIndex = false;
+    private String cachedSaasLandingHtml;
+    private boolean saasLandingExists = false;
 
     @PostConstruct
     public void init() {
@@ -49,6 +51,20 @@ public class ReactRoutingController {
         // Always initialize callback HTML (used for OAuth desktop flow)
         this.cachedCallbackHtml = buildCallbackHtml();
 
+        // SaaS landing page: only present on the classpath when the :saas module is bundled
+        // (app/saas/src/main/resources/static/saas-landing.html). When present it replaces the
+        // root page so the SaaS API host shows its own landing instead of the OSS API-only page.
+        ClassPathResource saasLanding = new ClassPathResource("static/saas-landing.html");
+        if (saasLanding.exists()) {
+            try (InputStream in = saasLanding.getInputStream()) {
+                this.cachedSaasLandingHtml = new String(in.readAllBytes(), StandardCharsets.UTF_8);
+                this.saasLandingExists = true;
+                log.info("SaaS landing page detected; serving it at '/' and '/index.html'");
+            } catch (Exception ex) {
+                log.warn("Failed to read saas-landing.html; falling back to index.html", ex);
+            }
+        }
+
         // Check for external index.html first (customFiles/static/)
         Path externalIndexPath = Paths.get(InstallationPathConfig.getStaticPath(), "index.html");
         log.debug("Checking for custom index.html at: {}", externalIndexPath);
@@ -132,6 +148,18 @@ public class ReactRoutingController {
     @GetMapping(
             value = {"/", "/index.html"},
             produces = MediaType.TEXT_HTML_VALUE)
+    public ResponseEntity<String> serveRootPage(HttpServletRequest request) {
+        // Swap ONLY the root page for SaaS. SPA entry points that delegate to serveIndexHtml
+        // (/auth/callback, /share/{token}, forwarded routes) keep serving the normal shell.
+        if (saasLandingExists && cachedSaasLandingHtml != null) {
+            return ResponseEntity.ok()
+                    .cacheControl(CacheControl.noCache().mustRevalidate())
+                    .contentType(MediaType.TEXT_HTML)
+                    .body(cachedSaasLandingHtml);
+        }
+        return serveIndexHtml(request);
+    }
+
     public ResponseEntity<String> serveIndexHtml(HttpServletRequest request) {
         try {
             if (indexHtmlExists && cachedIndexHtml != null) {

app/core/src/main/java/stirling/software/SPDF/exception/GlobalExceptionHandler.java

@@ -24,6 +24,7 @@ import org.springframework.web.multipart.MaxUploadSizeExceededException;
 import org.springframework.web.multipart.support.MissingServletRequestPartException;
 import org.springframework.web.server.ResponseStatusException;
 import org.springframework.web.servlet.NoHandlerFoundException;
+import org.springframework.web.servlet.resource.NoResourceFoundException;
 
 import jakarta.servlet.http.HttpServletRequest;
 import jakarta.servlet.http.HttpServletResponse;
@@ -993,6 +994,49 @@ public class GlobalExceptionHandler {
                 .body(problemDetail);
     }
 
+    /** Unmapped path → clean 404 instead of falling through to the generic 500 catch-all. */
+    @ExceptionHandler(NoResourceFoundException.class)
+    public ResponseEntity<ProblemDetail> handleNoResourceFound(
+            NoResourceFoundException ex, HttpServletRequest request) {
+        // /api/* miss = likely missing controller (operator-relevant); other paths = favicons,
+        // robots.txt, scanner noise. Demote the latter so prod logs aren't flooded.
+        String uri = request.getRequestURI();
+        if (uri != null && uri.startsWith("/api/")) {
+            log.warn("No resource at {}: {}", uri, ex.getMessage());
+        } else {
+            log.debug("No resource at {}: {}", uri, ex.getMessage());
+        }
+
+        String title = getLocalizedMessage("error.notFound.title", ErrorTitles.NOT_FOUND_DEFAULT);
+        String detail =
+                getLocalizedMessage(
+                        "error.notFound.detail",
+                        String.format(
+                                "No endpoint found for %s %s",
+                                request.getMethod(), request.getRequestURI()),
+                        request.getMethod(),
+                        request.getRequestURI());
+
+        ProblemDetail problemDetail =
+                createBaseProblemDetail(HttpStatus.NOT_FOUND, detail, request);
+        problemDetail.setType(URI.create(ErrorTypes.NOT_FOUND));
+        problemDetail.setTitle(title);
+        problemDetail.setProperty("title", title);
+        problemDetail.setProperty("method", request.getMethod());
+        addStandardHints(
+                problemDetail,
+                "error.notFound.hints",
+                List.of(
+                        "Verify the URL path and HTTP method are correct.",
+                        "Check the API base path and version if applicable.",
+                        "Ensure there are no typos in the endpoint path."));
+        problemDetail.setProperty("actionRequired", "Use a valid endpoint URL and method.");
+
+        return ResponseEntity.status(HttpStatus.NOT_FOUND)
+                .contentType(PROBLEM_JSON)
+                .body(problemDetail);
+    }
+
     /**
      * Handle IllegalArgumentException.
      *

app/core/src/main/java/stirling/software/SPDF/model/api/converters/ConvertPDFToMarkdown.java

@@ -1,11 +1,13 @@
 package stirling.software.SPDF.model.api.converters;
 
-import org.springframework.core.io.Resource;
+import java.nio.charset.StandardCharsets;
+
 import org.springframework.http.MediaType;
 import org.springframework.http.ResponseEntity;
 import org.springframework.web.bind.annotation.ModelAttribute;
 import org.springframework.web.multipart.MultipartFile;
 
+import io.github.pixee.security.Filenames;
 import io.swagger.v3.oas.annotations.Operation;
 
 import lombok.RequiredArgsConstructor;
@@ -15,8 +17,11 @@ import stirling.software.common.annotations.AutoJobPostMapping;
 import stirling.software.common.annotations.api.ConvertApi;
 import stirling.software.common.enumeration.ResourceWeight;
 import stirling.software.common.model.api.PDFFile;
-import stirling.software.common.util.PDFToFile;
+import stirling.software.common.pdf.PdfMarkdownConverter;
+import stirling.software.common.util.TempFile;
 import stirling.software.common.util.TempFileManager;
+import stirling.software.common.util.WebResponseUtils;
+import stirling.software.jpdfium.PdfDocument;
 
 @ConvertApi
 @RequiredArgsConstructor
@@ -33,10 +38,27 @@ public class ConvertPDFToMarkdown {
             summary = "Convert PDF to Markdown",
             description =
                     "This endpoint converts a PDF file to Markdown format. Input:PDF Output:Markdown Type:SISO")
-    public ResponseEntity<Resource> processPdfToMarkdown(@ModelAttribute PDFFile file)
+    public ResponseEntity<byte[]> processPdfToMarkdown(@ModelAttribute PDFFile file)
             throws Exception {
         MultipartFile inputFile = file.getFileInput();
-        PDFToFile pdfToFile = new PDFToFile(tempFileManager);
-        return pdfToFile.processPdfToMarkdown(inputFile);
+
+        String originalName = Filenames.toSimpleFileName(inputFile.getOriginalFilename());
+        String baseName =
+                originalName.contains(".")
+                        ? originalName.substring(0, originalName.lastIndexOf('.'))
+                        : originalName;
+
+        String markdown;
+        try (TempFile tempInput = new TempFile(tempFileManager, ".pdf")) {
+            inputFile.transferTo(tempInput.getFile());
+            try (PdfDocument doc = PdfDocument.open(tempInput.getPath())) {
+                markdown = new PdfMarkdownConverter().convert(doc);
+            }
+        }
+
+        return WebResponseUtils.bytesToWebResponse(
+                markdown.getBytes(StandardCharsets.UTF_8),
+                baseName + ".md",
+                MediaType.valueOf("text/markdown"));
     }
 }

app/core/src/main/java/stirling/software/SPDF/service/ApiDocService.java

@@ -63,6 +63,7 @@ public class ApiDocService implements stirling.software.common.service.ToolMetad
         return "http://localhost:" + port + contextPath + "/v1/api-docs";
     }
 
+    @Override
     public List<String> getExtensionTypes(boolean output, String operationName) {
         if (outputToFileTypes.isEmpty()) {
             outputToFileTypes.put("PDF", List.of("pdf"));

app/core/src/main/java/stirling/software/SPDF/utils/SvgOverlayUtil.java

@@ -10,6 +10,7 @@ import org.apache.batik.bridge.GVTBuilder;
 import org.apache.batik.bridge.UserAgent;
 import org.apache.batik.bridge.UserAgentAdapter;
 import org.apache.batik.gvt.GraphicsNode;
+import org.apache.batik.util.ParsedURL;
 import org.apache.batik.util.XMLResourceDescriptor;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
@@ -39,7 +40,16 @@ public class SvgOverlayUtil {
                 svgDoc = factory.createSVGDocument("file:///overlay.svg", inputStream);
             }
 
-            UserAgent userAgent = new UserAgentAdapter();
+            UserAgent userAgent =
+                    new UserAgentAdapter() {
+                        @Override
+                        public void checkLoadExternalResource(
+                                ParsedURL resourceURL, ParsedURL docURL) {
+                            throw new SecurityException(
+                                    "External resource loading is disabled for SVG overlays: "
+                                            + resourceURL);
+                        }
+                    };
             DocumentLoader loader = new DocumentLoader(userAgent);
             BridgeContext ctx = new BridgeContext(userAgent, loader);
             ctx.setDynamicState(BridgeContext.DYNAMIC);

app/core/src/main/resources/application.properties

@@ -93,6 +93,11 @@ posthog.host=https://eu.i.posthog.com
 
 spring.main.allow-bean-definition-overriding=true
 
+# spring-data-redis is on the classpath only for the optional Valkey backplane (which wires its own
+# factory); exclude Spring Boot's stock Redis auto-config so a default install doesn't create a dead
+# localhost:6379 factory that flips /actuator/health to DOWN.
+spring.autoconfigure.exclude=org.springframework.boot.data.redis.autoconfigure.DataRedisAutoConfiguration,org.springframework.boot.data.redis.autoconfigure.DataRedisReactiveAutoConfiguration
+
 # Set up a consistent temporary directory location
 java.io.tmpdir=${stirling.tempfiles.directory:${java.io.tmpdir}/stirling-pdf}
 

app/core/src/main/resources/settings.yml.template

@@ -364,6 +364,36 @@ aiEngine:
   url: http://localhost:5001 # URL of the Python AI engine
   timeoutSeconds: 120 # Timeout in seconds for AI engine requests
 
+policies:
+  # Folder automations can read from and write to the directories you allow here, so treat this as a
+  # security boundary. Leave allowedFolderRoots empty (default) to disable folder sources/outputs
+  # entirely; list absolute directories to permit folder access only within them. Stirling's own
+  # config directory is always off-limits, and folder access is always disabled in SaaS mode.
+  allowedFolderRoots: [] # e.g. ["/data/inbox", "/data/outbox"]
+  scheduleSweepSeconds: 60 # How often (seconds) scheduled policies are checked for being due
+  watchReconcileSeconds: 300 # How often (seconds) folder-watch re-syncs watches and re-runs as a safety net for missed events
+  watchQuietPeriodMs: 500 # How long (ms) folder-watch coalesces a burst of file events into a single run
+  streamTimeoutMs: 1800000 # SSE timeout (ms) for live run-progress streams
+  runExpiryMinutes: 30 # How long (minutes) a finished run's in-memory state is kept before eviction
+
+# Model Context Protocol (MCP) server. Exposes Stirling's PDF tools (grouped by namespace)
+# plus the AI agents to MCP clients (Inspector, Claude Desktop, custom). OAuth-protected.
+# Disabled by default - enable explicitly per deployment after configuring mcp.auth.
+mcp:
+  enabled: false # Master switch. 'false' (default) means no /mcp endpoint, no metadata, no beans wired.
+  scopesEnabled: true # Enforce mcp.tools.read / mcp.tools.write scopes derived from operation category
+  allowedOperations: [] # Tool allow-list (operation ids, e.g. ['compress-pdf']). Empty = all. When set, ONLY these are exposed over MCP.
+  blockedOperations: [] # Tool deny-list (operation ids). Always removed from MCP even if otherwise allowed.
+  auth:
+    mode: oauth # 'oauth' (full OAuth2 resource server) or 'apikey' (Stirling per-user API key via X-API-KEY header; no external IdP needed - the low-friction self-host option)
+    issuerUri: "" # OAuth2 issuer URI (e.g. http://localhost:9000). Required when mode=oauth.
+    jwksUri: "" # JWKS URI. Blank -> derived from issuer's /.well-known/openid-configuration.
+    resourceId: "" # RFC 8707 resource identifier of THIS MCP server (e.g. http://localhost:8080/mcp).
+                   # Required: tokens must list this id in `aud` or the request is rejected.
+    usernameClaim: sub # JWT claim matched against a Stirling username (e.g. 'sub', 'email', 'preferred_username')
+    requireExistingAccount: true # Reject tokens whose subject has no enabled Stirling account (recommended)
+  engineCapabilityRefreshMinutes: 5 # How often to refresh the AI capabilities manifest from the engine
+
 # Cluster configuration. NOT YET ENABLED - scaffolding for later work. Leave at defaults.
 cluster:
   enabled: false # Master switch. 'false' (default) wires the in-process backplane and skips all cluster checks. Single-instance installs do not need to change anything here.

app/core/src/test/java/stirling/software/SPDF/controller/api/RearrangePagesPDFControllerTest.java

@@ -4,10 +4,14 @@ import static org.junit.jupiter.api.Assertions.*;
 import static org.mockito.ArgumentMatchers.anyString;
 import static org.mockito.Mockito.*;
 
+import java.io.ByteArrayOutputStream;
 import java.io.File;
 import java.io.IOException;
 import java.nio.file.Files;
+import java.util.ArrayList;
+import java.util.List;
 
+import org.apache.pdfbox.Loader;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
 import org.junit.jupiter.api.BeforeEach;
@@ -56,6 +60,38 @@ class RearrangePagesPDFControllerTest {
                 "fileInput", "test.pdf", MediaType.APPLICATION_PDF_VALUE, new byte[] {1, 2, 3});
     }
 
+    /** Build a real, in-memory PDDocument with the requested number of blank pages. */
+    private PDDocument buildRealPdf(int pageCount) throws IOException {
+        PDDocument doc = new PDDocument();
+        for (int i = 0; i < pageCount; i++) {
+            doc.addPage(new PDPage());
+        }
+        return doc;
+    }
+
+    /**
+     * Returns the underlying {@link org.apache.pdfbox.cos.COSDictionary} for each page in document
+     * order. PDPageTree returns a fresh PDPage wrapper per get(), so comparing wrappers with
+     * assertSame is unreliable - the COSDictionary identity is the stable handle.
+     */
+    private List<Object> snapshotCosPages(PDDocument doc) {
+        List<Object> snapshot = new ArrayList<>();
+        for (PDPage p : doc.getPages()) {
+            snapshot.add(p.getCOSObject());
+        }
+        return snapshot;
+    }
+
+    private List<Object> reloadAndSnapshot(ResponseEntity<Resource> response) throws IOException {
+        try (var in = response.getBody().getInputStream();
+                var baos = new ByteArrayOutputStream()) {
+            in.transferTo(baos);
+            try (PDDocument out = Loader.loadPDF(baos.toByteArray())) {
+                return snapshotCosPages(out);
+            }
+        }
+    }
+
     @Test
     void testDeletePages_Success() throws IOException {
         MockMultipartFile file = createMockPdf();
@@ -83,27 +119,23 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("REVERSE_ORDER");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page0 = mock(PDPage.class);
-        PDPage page1 = mock(PDPage.class);
-        PDPage page2 = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(3)) {
+            List<Object> originals = snapshotCosPages(realDoc);
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(3);
-        when(mockDoc.getPage(0)).thenReturn(page0);
-        when(mockDoc.getPage(1)).thenReturn(page1);
-        when(mockDoc.getPage(2)).thenReturn(page2);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        assertEquals(200, response.getStatusCode().value());
-        verify(mockNewDoc).addPage(page2);
-        verify(mockNewDoc).addPage(page1);
-        verify(mockNewDoc).addPage(page0);
+            assertNotNull(response);
+            assertEquals(200, response.getStatusCode().value());
+            List<Object> finalOrder = reloadAndSnapshot(response);
+            assertEquals(3, finalOrder.size());
+            // We can no longer compare references after a save/reload, so compare via
+            // the in-memory snapshot taken *after* the controller mutated the source.
+            List<Object> mutatedSource = snapshotCosPages(realDoc);
+            assertSame(originals.get(2), mutatedSource.get(0));
+            assertSame(originals.get(1), mutatedSource.get(1));
+            assertSame(originals.get(0), mutatedSource.get(2));
+        }
     }
 
     @Test
@@ -114,25 +146,18 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("REMOVE_FIRST");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page0 = mock(PDPage.class);
-        PDPage page1 = mock(PDPage.class);
-        PDPage page2 = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(3)) {
+            List<Object> originals = snapshotCosPages(realDoc);
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(3);
-        when(mockDoc.getPage(1)).thenReturn(page1);
-        when(mockDoc.getPage(2)).thenReturn(page2);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        verify(mockNewDoc).addPage(page1);
-        verify(mockNewDoc).addPage(page2);
-        verify(mockNewDoc, never()).addPage(page0);
+            assertNotNull(response);
+            List<Object> mutated = snapshotCosPages(realDoc);
+            assertEquals(2, mutated.size());
+            assertSame(originals.get(1), mutated.get(0));
+            assertSame(originals.get(2), mutated.get(1));
+        }
     }
 
     @Test
@@ -143,23 +168,18 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("REMOVE_LAST");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page0 = mock(PDPage.class);
-        PDPage page1 = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(3)) {
+            List<Object> originals = snapshotCosPages(realDoc);
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(3);
-        when(mockDoc.getPage(0)).thenReturn(page0);
-        when(mockDoc.getPage(1)).thenReturn(page1);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        verify(mockNewDoc).addPage(page0);
-        verify(mockNewDoc).addPage(page1);
+            assertNotNull(response);
+            List<Object> mutated = snapshotCosPages(realDoc);
+            assertEquals(2, mutated.size());
+            assertSame(originals.get(0), mutated.get(0));
+            assertSame(originals.get(1), mutated.get(1));
+        }
     }
 
     @Test
@@ -170,21 +190,19 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("REMOVE_FIRST_AND_LAST");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page1 = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(4)) {
+            List<Object> originals = snapshotCosPages(realDoc);
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(4);
-        when(mockDoc.getPage(1)).thenReturn(page1);
-        when(mockDoc.getPage(2)).thenReturn(page1);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        assertEquals(200, response.getStatusCode().value());
+            assertNotNull(response);
+            assertEquals(200, response.getStatusCode().value());
+            List<Object> mutated = snapshotCosPages(realDoc);
+            assertEquals(2, mutated.size());
+            assertSame(originals.get(1), mutated.get(0));
+            assertSame(originals.get(2), mutated.get(1));
+        }
     }
 
     @Test
@@ -195,23 +213,15 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("DUPLEX_SORT");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page0 = mock(PDPage.class);
-        PDPage page1 = mock(PDPage.class);
-        PDPage page2 = mock(PDPage.class);
-        PDPage page3 = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(4)) {
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(4);
-        when(mockDoc.getPage(anyInt())).thenReturn(page0);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        assertEquals(200, response.getStatusCode().value());
+            assertNotNull(response);
+            assertEquals(200, response.getStatusCode().value());
+            assertEquals(4, realDoc.getNumberOfPages());
+        }
     }
 
     @Test
@@ -222,20 +232,15 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("BOOKLET_SORT");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(4)) {
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(4);
-        when(mockDoc.getPage(anyInt())).thenReturn(page);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        assertEquals(200, response.getStatusCode().value());
+            assertNotNull(response);
+            assertEquals(200, response.getStatusCode().value());
+            assertEquals(4, realDoc.getNumberOfPages());
+        }
     }
 
     @Test
@@ -246,20 +251,15 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("ODD_EVEN_SPLIT");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(4)) {
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(4);
-        when(mockDoc.getPage(anyInt())).thenReturn(page);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        assertEquals(200, response.getStatusCode().value());
+            assertNotNull(response);
+            assertEquals(200, response.getStatusCode().value());
+            assertEquals(4, realDoc.getNumberOfPages());
+        }
     }
 
     @Test
@@ -270,24 +270,20 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("3,1,2");
         request.setCustomMode("custom");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page0 = mock(PDPage.class);
-        PDPage page1 = mock(PDPage.class);
-        PDPage page2 = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(3)) {
+            List<Object> originals = snapshotCosPages(realDoc);
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(3);
-        when(mockDoc.getPage(0)).thenReturn(page0);
-        when(mockDoc.getPage(1)).thenReturn(page1);
-        when(mockDoc.getPage(2)).thenReturn(page2);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        assertEquals(200, response.getStatusCode().value());
+            assertNotNull(response);
+            assertEquals(200, response.getStatusCode().value());
+            List<Object> mutated = snapshotCosPages(realDoc);
+            assertEquals(3, mutated.size());
+            assertSame(originals.get(2), mutated.get(0));
+            assertSame(originals.get(0), mutated.get(1));
+            assertSame(originals.get(1), mutated.get(2));
+        }
     }
 
     @Test
@@ -298,21 +294,15 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("3");
         request.setCustomMode("DUPLICATE");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(2)) {
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(2);
-        when(mockDoc.getPage(anyInt())).thenReturn(page);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        // 2 pages * 3 duplicates = 6 addPage calls
-        verify(mockNewDoc, times(6)).addPage(page);
+            assertNotNull(response);
+            // 2 pages * 3 duplicates = 6 final pages
+            assertEquals(6, realDoc.getNumberOfPages());
+        }
     }
 
     @Test
@@ -323,19 +313,14 @@ class RearrangePagesPDFControllerTest {
         request.setPageNumbers("");
         request.setCustomMode("SIDE_STITCH_BOOKLET_SORT");
 
-        PDDocument mockDoc = mock(PDDocument.class);
-        PDDocument mockNewDoc = mock(PDDocument.class);
-        PDPage page = mock(PDPage.class);
+        try (PDDocument realDoc = buildRealPdf(4)) {
+            when(pdfDocumentFactory.load(file)).thenReturn(realDoc);
 
-        when(pdfDocumentFactory.load(file)).thenReturn(mockDoc);
-        when(mockDoc.getNumberOfPages()).thenReturn(4);
-        when(mockDoc.getPage(anyInt())).thenReturn(page);
-        when(pdfDocumentFactory.createNewDocumentBasedOnOldDocument(mockDoc))
-                .thenReturn(mockNewDoc);
+            ResponseEntity<Resource> response = controller.rearrangePages(request);
 
-        ResponseEntity<Resource> response = controller.rearrangePages(request);
-
-        assertNotNull(response);
-        assertEquals(200, response.getStatusCode().value());
+            assertNotNull(response);
+            assertEquals(200, response.getStatusCode().value());
+            assertEquals(4, realDoc.getNumberOfPages());
+        }
     }
 }

app/core/src/test/java/stirling/software/SPDF/controller/api/misc/OverlayImageControllerTest.java

@@ -32,6 +32,7 @@ import org.springframework.mock.web.MockMultipartFile;
 
 import stirling.software.SPDF.model.api.misc.OverlayImageRequest;
 import stirling.software.common.service.CustomPDFDocumentFactory;
+import stirling.software.common.util.SvgSanitizer;
 import stirling.software.common.util.TempFile;
 import stirling.software.common.util.TempFileManager;
 import stirling.software.common.util.WebResponseUtils;
@@ -52,6 +53,7 @@ class OverlayImageControllerTest {
 
     @Mock private CustomPDFDocumentFactory pdfDocumentFactory;
     @Mock private TempFileManager tempFileManager;
+    @Mock private SvgSanitizer svgSanitizer;
 
     @InjectMocks private OverlayImageController controller;
 
@@ -205,6 +207,52 @@ class OverlayImageControllerTest {
         mockDoc.close();
     }
 
+    @Test
+    void overlayImage_svgInput_sanitizedBeforeOverlay() throws Exception {
+        byte[] maliciousSvg =
+                ("<svg xmlns=\"http://www.w3.org/2000/svg\""
+                                + " xmlns:xlink=\"http://www.w3.org/1999/xlink\""
+                                + " width=\"10\" height=\"10\">"
+                                + "<image x=\"0\" y=\"0\" width=\"10\" height=\"10\""
+                                + " xlink:href=\"file:///etc/passwd\"/>"
+                                + "</svg>")
+                        .getBytes();
+        byte[] sanitized =
+                ("<svg xmlns=\"http://www.w3.org/2000/svg\" width=\"10\" height=\"10\">"
+                                + "<image x=\"0\" y=\"0\" width=\"10\" height=\"10\"/>"
+                                + "</svg>")
+                        .getBytes();
+        when(svgSanitizer.sanitize(maliciousSvg)).thenReturn(sanitized);
+
+        MockMultipartFile svgFile =
+                new MockMultipartFile("imageFile", "overlay.svg", "image/svg+xml", maliciousSvg);
+        OverlayImageRequest request = new OverlayImageRequest();
+        request.setFileInput(pdfFile);
+        request.setImageFile(svgFile);
+        request.setX(0);
+        request.setY(0);
+        request.setEveryPage(false);
+
+        PDDocument mockDoc = new PDDocument();
+        mockDoc.addPage(new PDPage(PDRectangle.A4));
+        when(pdfDocumentFactory.load(any(byte[].class))).thenReturn(mockDoc);
+
+        try (MockedStatic<WebResponseUtils> mockedWebResponse =
+                mockStatic(WebResponseUtils.class)) {
+            mockedWebResponse
+                    .when(
+                            () ->
+                                    WebResponseUtils.pdfFileToWebResponse(
+                                            any(TempFile.class), anyString()))
+                    .thenReturn(streamingOk("result".getBytes()));
+
+            controller.overlayImage(request);
+        }
+        mockDoc.close();
+
+        verify(svgSanitizer).sanitize(maliciousSvg);
+    }
+
     @Test
     void overlayImage_withCoordinates_usesXY() throws Exception {
         OverlayImageRequest request = new OverlayImageRequest();

app/core/src/test/java/stirling/software/SPDF/exception/GlobalExceptionHandlerTest.java

@@ -18,6 +18,7 @@ import org.mockito.Mock;
 import org.mockito.junit.jupiter.MockitoExtension;
 import org.springframework.context.MessageSource;
 import org.springframework.core.env.Environment;
+import org.springframework.http.HttpMethod;
 import org.springframework.http.HttpStatus;
 import org.springframework.http.ProblemDetail;
 import org.springframework.http.ResponseEntity;
@@ -27,6 +28,7 @@ import org.springframework.web.bind.MissingServletRequestParameterException;
 import org.springframework.web.multipart.MaxUploadSizeExceededException;
 import org.springframework.web.multipart.support.MissingServletRequestPartException;
 import org.springframework.web.servlet.NoHandlerFoundException;
+import org.springframework.web.servlet.resource.NoResourceFoundException;
 
 import jakarta.servlet.http.HttpServletRequest;
 import jakarta.servlet.http.HttpServletResponse;
@@ -208,6 +210,19 @@ class GlobalExceptionHandlerTest {
         assertEquals(HttpStatus.NOT_FOUND, resp.getStatusCode());
     }
 
+    // ---- NoResourceFoundException ----
+    // Regression guard: was falling through to the 500 catch-all.
+
+    @Test
+    void handleNoResourceFound_returns_404_not_500() {
+        when(request.getMethod()).thenReturn("GET");
+        NoResourceFoundException ex =
+                new NoResourceFoundException(HttpMethod.GET, "/api/v1/storage/folders", "");
+        ResponseEntity<ProblemDetail> resp = handler.handleNoResourceFound(ex, request);
+        assertEquals(HttpStatus.NOT_FOUND, resp.getStatusCode());
+        assertEquals("GET", resp.getBody().getProperties().get("method"));
+    }
+
     // ---- IllegalArgumentException ----
 
     @Test

app/core/src/test/java/stirling/software/SPDF/model/api/converters/ConvertPDFToMarkdownTest.java

@@ -1,16 +1,17 @@
 package stirling.software.SPDF.model.api.converters;
 
-import static org.junit.jupiter.api.Assertions.assertEquals;
 import static org.mockito.ArgumentMatchers.any;
 import static org.mockito.Mockito.*;
 import static org.springframework.test.web.servlet.request.MockMvcRequestBuilders.multipart;
 import static org.springframework.test.web.servlet.result.MockMvcResultMatchers.*;
 
+import java.io.File;
 import java.nio.charset.StandardCharsets;
+import java.nio.file.Path;
 
 import org.junit.jupiter.api.Test;
-import org.mockito.ArgumentCaptor;
 import org.mockito.MockedConstruction;
+import org.mockito.MockedStatic;
 import org.mockito.Mockito;
 import org.springframework.core.io.ByteArrayResource;
 import org.springframework.core.io.Resource;
@@ -21,9 +22,10 @@ import org.springframework.test.web.servlet.MockMvc;
 import org.springframework.test.web.servlet.setup.MockMvcBuilders;
 import org.springframework.web.bind.annotation.ExceptionHandler;
 import org.springframework.web.bind.annotation.RestControllerAdvice;
-import org.springframework.web.multipart.MultipartFile;
 
-import stirling.software.common.util.PDFToFile;
+import stirling.software.common.pdf.PdfMarkdownConverter;
+import stirling.software.common.util.TempFile;
+import stirling.software.jpdfium.PdfDocument;
 
 class ConvertPDFToMarkdownTest {
 
@@ -47,68 +49,68 @@ class ConvertPDFToMarkdownTest {
     @Test
     void pdfToMarkdownReturnsMarkdownBytes() throws Exception {
         byte[] md = "# heading\n\ncontent\n".getBytes(StandardCharsets.UTF_8);
+        String expectedMd = "# heading\n\ncontent\n";
 
-        try (MockedConstruction<PDFToFile> construction =
-                Mockito.mockConstruction(
-                        PDFToFile.class,
-                        (mock, ctx) -> {
-                            when(mock.processPdfToMarkdown(any(MultipartFile.class)))
-                                    .thenAnswer(
-                                            inv ->
-                                                    ResponseEntity.ok()
-                                                            .header("Content-Type", "text/markdown")
-                                                            .body(new ByteArrayResource(md)));
-                        })) {
+        File tmpFile = File.createTempFile("test", ".pdf");
+        tmpFile.deleteOnExit();
 
-            MockMvc mvc = mockMvc();
+        try (MockedConstruction<TempFile> tempMock =
+                        Mockito.mockConstruction(
+                                TempFile.class,
+                                (mock, ctx) -> {
+                                    when(mock.getFile()).thenReturn(tmpFile);
+                                    when(mock.getPath()).thenReturn(tmpFile.toPath());
+                                });
+                MockedStatic<PdfDocument> docStatic = Mockito.mockStatic(PdfDocument.class);
+                MockedConstruction<PdfMarkdownConverter> converterMock =
+                        Mockito.mockConstruction(
+                                PdfMarkdownConverter.class,
+                                (mock, ctx) -> when(mock.convert(any())).thenReturn(expectedMd))) {
+
+            PdfDocument mockDoc = Mockito.mock(PdfDocument.class);
+            docStatic.when(() -> PdfDocument.open(any(Path.class))).thenReturn(mockDoc);
 
             MockMultipartFile file =
                     new MockMultipartFile(
-                            "fileInput", // must match the field name in PDFFile
-                            "input.pdf",
-                            "application/pdf",
-                            new byte[] {1, 2, 3});
+                            "fileInput", "input.pdf", "application/pdf", new byte[] {1, 2, 3});
 
-            // ResponseEntity<Resource> is written synchronously on the request thread,
-            // so there is no async dispatch to wait for (unlike the old StreamingResponseBody
-            // path).
-            mvc.perform(multipart("/api/v1/convert/pdf/markdown").file(file))
+            mockMvc()
+                    .perform(multipart("/api/v1/convert/pdf/markdown").file(file))
                     .andExpect(status().isOk())
                     .andExpect(header().string("Content-Type", "text/markdown"))
                     .andExpect(content().bytes(md));
-
-            // Verify that exactly one instance was created
-            assert construction.constructed().size() == 1;
-
-            // And that the uploaded file was passed to processPdfToMarkdown()
-            PDFToFile created = construction.constructed().get(0);
-            ArgumentCaptor<MultipartFile> captor = ArgumentCaptor.forClass(MultipartFile.class);
-            verify(created, times(1)).processPdfToMarkdown(captor.capture());
-            MultipartFile passed = captor.getValue();
-
-            // Minimal plausibility checks
-            assertEquals("input.pdf", passed.getOriginalFilename());
-            assertEquals("application/pdf", passed.getContentType());
         }
     }
 
     @Test
     void pdfToMarkdownWhenServiceThrowsReturns500() throws Exception {
-        try (MockedConstruction<PDFToFile> ignored =
-                Mockito.mockConstruction(
-                        PDFToFile.class,
-                        (mock, ctx) -> {
-                            when(mock.processPdfToMarkdown(any(MultipartFile.class)))
-                                    .thenThrow(new RuntimeException("boom"));
-                        })) {
+        File tmpFile = File.createTempFile("test", ".pdf");
+        tmpFile.deleteOnExit();
 
-            MockMvc mvc = mockMvc();
+        try (MockedConstruction<TempFile> tempMock =
+                        Mockito.mockConstruction(
+                                TempFile.class,
+                                (mock, ctx) -> {
+                                    when(mock.getFile()).thenReturn(tmpFile);
+                                    when(mock.getPath()).thenReturn(tmpFile.toPath());
+                                });
+                MockedStatic<PdfDocument> docStatic = Mockito.mockStatic(PdfDocument.class);
+                MockedConstruction<PdfMarkdownConverter> converterMock =
+                        Mockito.mockConstruction(
+                                PdfMarkdownConverter.class,
+                                (mock, ctx) ->
+                                        when(mock.convert(any()))
+                                                .thenThrow(new RuntimeException("boom")))) {
+
+            PdfDocument mockDoc = Mockito.mock(PdfDocument.class);
+            docStatic.when(() -> PdfDocument.open(any(Path.class))).thenReturn(mockDoc);
 
             MockMultipartFile file =
                     new MockMultipartFile(
                             "fileInput", "x.pdf", "application/pdf", new byte[] {0x01});
 
-            mvc.perform(multipart("/api/v1/convert/pdf/markdown").file(file))
+            mockMvc()
+                    .perform(multipart("/api/v1/convert/pdf/markdown").file(file))
                     .andExpect(status().isInternalServerError());
         }
     }

app/proprietary/build.gradle

@@ -52,6 +52,10 @@ dependencies {
     api 'org.springframework.boot:spring-boot-starter-security'
     api 'org.springframework.boot:spring-boot-starter-data-jpa'
     api 'org.springframework.boot:spring-boot-starter-security-oauth2-client'
+    // MCP server (RFC 8707 audience binding + RFC 9728 metadata) - resource-server side only.
+    // Brings nimbus-jose-jwt onto the proprietary classpath if not already transitive via
+    // oauth2-client; on Boot 4.0.6 the delta is around 130KB because nimbus is already pulled.
+    api 'org.springframework.boot:spring-boot-starter-oauth2-resource-server'
     api 'org.springframework.boot:spring-boot-starter-mail'
     api 'org.springframework.boot:spring-boot-starter-cache'
     api 'com.github.ben-manes.caffeine:caffeine'

app/proprietary/src/main/java/stirling/software/proprietary/cluster/s3/S3FileStore.java

@@ -6,6 +6,8 @@ import java.io.InputStream;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.nio.file.StandardCopyOption;
+import java.util.Collections;
+import java.util.Map;
 import java.util.Optional;
 import java.util.UUID;
 
@@ -35,6 +37,7 @@ import software.amazon.awssdk.services.s3.model.S3Exception;
 public class S3FileStore implements FileStore, AutoCloseable {
 
     public static final String DEFAULT_KEY_PREFIX = "transient/";
+    static final String OWNER_METADATA_KEY = "owner";
 
     private final S3Client s3Client;
     private final String bucket;
@@ -64,7 +67,7 @@ public class S3FileStore implements FileStore, AutoCloseable {
     }
 
     @Override
-    public Stored store(InputStream in, String originalName) throws IOException {
+    public Stored store(InputStream in, String originalName, String owner) throws IOException {
         String fileId = UUID.randomUUID().toString();
         // S3 PUT requires a known content-length; spool to a temp file first so memory stays
         // bounded for large payloads, then stream the file to S3 via RequestBody.fromFile.
@@ -75,10 +78,13 @@ public class S3FileStore implements FileStore, AutoCloseable {
                 Files.copy(src, tempFile, StandardCopyOption.REPLACE_EXISTING);
             }
             size = Files.size(tempFile);
-            PutObjectRequest request =
-                    PutObjectRequest.builder().bucket(bucket).key(resolveKey(fileId)).build();
+            PutObjectRequest.Builder builder =
+                    PutObjectRequest.builder().bucket(bucket).key(resolveKey(fileId));
+            if (owner != null && !owner.isBlank()) {
+                builder.metadata(Map.of(OWNER_METADATA_KEY, owner));
+            }
             try {
-                s3Client.putObject(request, RequestBody.fromFile(tempFile));
+                s3Client.putObject(builder.build(), RequestBody.fromFile(tempFile));
             } catch (SdkException e) {
                 throw new IOException("Failed to upload object to S3", e);
             }
@@ -185,6 +191,36 @@ public class S3FileStore implements FileStore, AutoCloseable {
         }
     }
 
+    @Override
+    public String getOwner(String fileId) throws IOException {
+        try {
+            validateFileId(fileId);
+        } catch (IllegalArgumentException e) {
+            return null;
+        }
+        HeadObjectRequest request =
+                HeadObjectRequest.builder().bucket(bucket).key(resolveKey(fileId)).build();
+        try {
+            HeadObjectResponse response = s3Client.headObject(request);
+            Map<String, String> metadata =
+                    Optional.ofNullable(response.metadata()).orElse(Collections.emptyMap());
+            String owner = metadata.get(OWNER_METADATA_KEY);
+            if (owner != null && !owner.isBlank()) {
+                return owner;
+            }
+            return null;
+        } catch (NoSuchKeyException e) {
+            return null;
+        } catch (S3Exception e) {
+            if (e.statusCode() == 404) {
+                return null;
+            }
+            throw new IOException("Failed to read owner metadata from S3", e);
+        } catch (SdkException e) {
+            throw new IOException("Failed to read owner metadata from S3", e);
+        }
+    }
+
     @Override
     public void close() {
         if (!ownsClient) {
@@ -205,7 +241,7 @@ public class S3FileStore implements FileStore, AutoCloseable {
         if (fileId == null || fileId.isBlank()) {
             throw new IllegalArgumentException("File ID must not be blank");
         }
-        if (fileId.contains("..") || fileId.contains("/") || fileId.contains("\\")) {
+        if (fileId.contains(".") || fileId.contains("/") || fileId.contains("\\")) {
             throw new IllegalArgumentException("Invalid file ID");
         }
     }

app/proprietary/src/main/java/stirling/software/proprietary/mcp/McpCallContext.java

@@ -0,0 +1,15 @@
+package stirling.software.proprietary.mcp;
+
+import java.util.Set;
+
+/** Per-call context: resolved Stirling identity and granted scopes for an {@link McpTool#call}. */
+public record McpCallContext(
+        String stirlingUserId, Set<String> grantedScopes, boolean scopesEnabled) {
+
+    public boolean hasScope(String required) {
+        if (!scopesEnabled) {
+            return true;
+        }
+        return required == null || required.isBlank() || grantedScopes.contains(required);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/McpServerController.java

@@ -0,0 +1,217 @@
+package stirling.software.proprietary.mcp;
+
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.http.HttpStatus;
+import org.springframework.http.MediaType;
+import org.springframework.http.ResponseEntity;
+import org.springframework.http.converter.HttpMessageNotReadableException;
+import org.springframework.web.bind.annotation.ExceptionHandler;
+import org.springframework.web.bind.annotation.PostMapping;
+import org.springframework.web.bind.annotation.RequestBody;
+import org.springframework.web.bind.annotation.RequestMapping;
+import org.springframework.web.bind.annotation.RestController;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.proprietary.mcp.jsonrpc.JsonRpcError;
+import stirling.software.proprietary.mcp.jsonrpc.JsonRpcRequest;
+import stirling.software.proprietary.mcp.jsonrpc.JsonRpcResponse;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ArrayNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/** Streamable-HTTP MCP server endpoint serving JSON-RPC 2.0 frames on {@code POST /mcp}. */
+@Slf4j
+@RestController
+@RequestMapping
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class McpServerController {
+
+    private static final String PREFERRED_PROTOCOL_VERSION = "2025-06-18";
+    private static final Set<String> SUPPORTED_PROTOCOL_VERSIONS =
+            Set.of("2025-06-18", "2025-03-26", "2024-11-05");
+    private static final String SERVER_NAME = "stirling-pdf-mcp";
+
+    private final ObjectMapper mapper;
+    private final ApplicationProperties applicationProperties;
+    private final Map<String, McpTool> toolsByName;
+
+    public McpServerController(
+            ObjectMapper mapper, ApplicationProperties applicationProperties, List<McpTool> tools) {
+        this.mapper = mapper;
+        this.applicationProperties = applicationProperties;
+        this.toolsByName = new HashMap<>();
+        for (McpTool tool : tools) {
+            this.toolsByName.put(tool.name(), tool);
+        }
+        log.info(
+                "MCP server controller wired with {} tool(s): {}",
+                toolsByName.size(),
+                toolsByName.keySet());
+    }
+
+    @PostMapping(
+            path = "/mcp",
+            consumes = MediaType.APPLICATION_JSON_VALUE,
+            produces = MediaType.APPLICATION_JSON_VALUE)
+    public ResponseEntity<?> handle(@RequestBody JsonNode body) {
+        JsonRpcRequest request = decode(body);
+        if (request == null) {
+            // Valid JSON but not a JSON-RPC request -> Invalid Request, not Parse error.
+            return ResponseEntity.badRequest()
+                    .body(
+                            JsonRpcResponse.failure(
+                                    null,
+                                    JsonRpcError.invalidRequest(
+                                            "Body is not a valid JSON-RPC 2.0 request")));
+        }
+        if (request.isNotification()) {
+            log.debug("Notification received: {}", sanitizeForLog(request.method()));
+            return ResponseEntity.status(HttpStatus.NO_CONTENT).build();
+        }
+        JsonRpcResponse response;
+        try {
+            response = dispatch(request);
+        } catch (RuntimeException e) {
+            log.warn(
+                    "MCP dispatch failed for method {}: {}",
+                    sanitizeForLog(request.method()),
+                    e.getMessage(),
+                    e);
+            response =
+                    JsonRpcResponse.failure(
+                            request.id(),
+                            JsonRpcError.internalError(
+                                    "Internal error handling " + request.method()));
+        }
+        return ResponseEntity.ok(response);
+    }
+
+    /** Wrap malformed-JSON failures (caught before {@link #handle}) as a JSON-RPC Parse error. */
+    @ExceptionHandler(HttpMessageNotReadableException.class)
+    public ResponseEntity<JsonRpcResponse> handleUnreadable(HttpMessageNotReadableException ex) {
+        return ResponseEntity.badRequest()
+                .contentType(MediaType.APPLICATION_JSON)
+                .body(
+                        JsonRpcResponse.failure(
+                                null, JsonRpcError.parseError("Request body is not valid JSON")));
+    }
+
+    private static String sanitizeForLog(String value) {
+        return value == null ? null : value.replace('\r', ' ').replace('\n', ' ');
+    }
+
+    private JsonRpcRequest decode(JsonNode body) {
+        if (body == null || !body.isObject()) {
+            return null;
+        }
+        JsonNode jsonrpc = body.get("jsonrpc");
+        JsonNode method = body.get("method");
+        if (jsonrpc == null || !"2.0".equals(jsonrpc.asText())) {
+            return null;
+        }
+        if (method == null || !method.isTextual()) {
+            return null;
+        }
+        return new JsonRpcRequest(
+                jsonrpc.asText(), body.get("id"), method.asText(), body.get("params"));
+    }
+
+    private JsonRpcResponse dispatch(JsonRpcRequest request) {
+        return switch (request.method()) {
+            case "initialize" ->
+                    JsonRpcResponse.success(request.id(), initializeResult(request.params()));
+            case "tools/list" -> JsonRpcResponse.success(request.id(), toolsListResult());
+            case "tools/call" -> handleToolsCall(request);
+            case "ping" -> JsonRpcResponse.success(request.id(), mapper.createObjectNode());
+            case "notifications/initialized" ->
+                    JsonRpcResponse.success(request.id(), mapper.createObjectNode());
+            default ->
+                    JsonRpcResponse.failure(
+                            request.id(), JsonRpcError.methodNotFound(request.method()));
+        };
+    }
+
+    private ObjectNode initializeResult(JsonNode params) {
+        ObjectNode result = mapper.createObjectNode();
+        // Echo the client's requested protocolVersion when supported, else advertise our preferred.
+        String requested =
+                params != null && params.hasNonNull("protocolVersion")
+                        ? params.get("protocolVersion").asText()
+                        : null;
+        String negotiated =
+                requested != null && SUPPORTED_PROTOCOL_VERSIONS.contains(requested)
+                        ? requested
+                        : PREFERRED_PROTOCOL_VERSION;
+        result.put("protocolVersion", negotiated);
+        ObjectNode caps = result.putObject("capabilities");
+        caps.putObject("tools");
+        ObjectNode info = result.putObject("serverInfo");
+        info.put("name", SERVER_NAME);
+        info.put("version", applicationProperties.getAutomaticallyGenerated().getAppVersion());
+        return result;
+    }
+
+    private ObjectNode toolsListResult() {
+        ObjectNode result = mapper.createObjectNode();
+        ArrayNode tools = result.putArray("tools");
+        for (McpTool t : toolsByName.values()) {
+            ObjectNode entry = mapper.createObjectNode();
+            entry.put("name", t.name());
+            entry.put("description", t.description());
+            entry.set("inputSchema", t.inputSchema());
+            tools.add(entry);
+        }
+        return result;
+    }
+
+    private JsonRpcResponse handleToolsCall(JsonRpcRequest request) {
+        JsonNode params = request.params();
+        if (params == null || !params.isObject()) {
+            return JsonRpcResponse.failure(
+                    request.id(), JsonRpcError.invalidParams("Missing params for tools/call"));
+        }
+        JsonNode nameNode = params.get("name");
+        if (nameNode == null || !nameNode.isTextual()) {
+            return JsonRpcResponse.failure(
+                    request.id(), JsonRpcError.invalidParams("Missing tool name"));
+        }
+        McpTool tool = toolsByName.get(nameNode.asText());
+        if (tool == null) {
+            return JsonRpcResponse.failure(
+                    request.id(), JsonRpcError.invalidParams("Unknown tool: " + nameNode.asText()));
+        }
+        JsonNode args = params.get("arguments");
+        McpCallContext context = resolveContext();
+        ObjectNode toolResult = tool.call(args == null ? mapper.createObjectNode() : args, context);
+        return JsonRpcResponse.success(request.id(), toolResult);
+    }
+
+    private McpCallContext resolveContext() {
+        boolean scopesEnabled = applicationProperties.getMcp().isScopesEnabled();
+        org.springframework.security.core.Authentication auth =
+                org.springframework.security.core.context.SecurityContextHolder.getContext()
+                        .getAuthentication();
+        // Fail closed: no/unauthenticated principal yields an empty context so scoped ops are
+        // refused.
+        if (auth == null || !auth.isAuthenticated() || auth.getName() == null) {
+            return new McpCallContext(null, Set.of(), scopesEnabled);
+        }
+        java.util.Set<String> scopes = new java.util.HashSet<>();
+        for (org.springframework.security.core.GrantedAuthority ga : auth.getAuthorities()) {
+            String authority = ga.getAuthority();
+            if (authority != null && authority.startsWith("SCOPE_")) {
+                scopes.add(authority.substring("SCOPE_".length()));
+            }
+        }
+        return new McpCallContext(auth.getName(), scopes, scopesEnabled);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/McpTool.java

@@ -0,0 +1,18 @@
+package stirling.software.proprietary.mcp;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/** Contract every MCP tool registered with the server must satisfy. */
+public interface McpTool {
+
+    String name();
+
+    String description();
+
+    /** The tool's {@code inputSchema} (an object JSON Schema) published in {@code tools/list}. */
+    ObjectNode inputSchema();
+
+    /** Execute the tool; the controller wraps any thrown exception as an MCP internal error. */
+    ObjectNode call(JsonNode arguments, McpCallContext context);
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/catalog/McpToolCatalog.java

@@ -0,0 +1,266 @@
+package stirling.software.proprietary.mcp.catalog;
+
+import java.lang.reflect.Method;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.LinkedHashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.Optional;
+import java.util.Set;
+import java.util.TreeSet;
+import java.util.concurrent.ConcurrentHashMap;
+
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.context.ApplicationContext;
+import org.springframework.context.event.ContextRefreshedEvent;
+import org.springframework.context.event.EventListener;
+import org.springframework.core.MethodParameter;
+import org.springframework.stereotype.Component;
+import org.springframework.web.bind.annotation.RequestMethod;
+import org.springframework.web.method.HandlerMethod;
+import org.springframework.web.servlet.mvc.method.RequestMappingInfo;
+import org.springframework.web.servlet.mvc.method.annotation.RequestMappingHandlerMapping;
+
+import io.swagger.v3.oas.annotations.Operation;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.SPDF.config.EndpointConfiguration;
+import stirling.software.common.model.ApplicationProperties;
+
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Discovers MCP-exposable operations and caches a per-op {@link OperationMeta}. Refreshed on {@link
+ * ContextRefreshedEvent} and filtered on read by {@link
+ * EndpointConfiguration#isEndpointEnabledForUri}. AI capabilities are fed in via {@link
+ * #replaceAiCapabilities}.
+ */
+@Slf4j
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class McpToolCatalog {
+
+    private static final String WRITE_SCOPE = "mcp.tools.write";
+
+    private final ApplicationContext applicationContext;
+    private final EndpointConfiguration endpointConfiguration;
+    private final ApplicationProperties applicationProperties;
+    private final SimpleSchemaGenerator schemaGenerator;
+    private final ObjectMapper objectMapper;
+
+    // Concurrent: written on the boot thread, read on request threads, AI map replaced at runtime.
+    private final Map<String, OperationMeta> pdfOps = new ConcurrentHashMap<>();
+
+    // Engine-driven AI capabilities. Replaced wholesale by the scheduled refresh task on a
+    // background thread while request threads read via findByOperationId/enabledOps. The volatile
+    // reference makes the swap publication-safe; readers either see the old or the new snapshot,
+    // never a partially-merged one.
+    private volatile Map<String, OperationMeta> aiOps = new ConcurrentHashMap<>();
+
+    public McpToolCatalog(
+            ApplicationContext applicationContext,
+            EndpointConfiguration endpointConfiguration,
+            ApplicationProperties applicationProperties,
+            ObjectMapper objectMapper) {
+        this.applicationContext = applicationContext;
+        this.endpointConfiguration = endpointConfiguration;
+        this.applicationProperties = applicationProperties;
+        this.schemaGenerator = new SimpleSchemaGenerator(objectMapper);
+        this.objectMapper = objectMapper;
+    }
+
+    /** Admin tool filter: non-empty allow list is a whitelist; block list always removes. */
+    private boolean isOperationAllowed(String id) {
+        ApplicationProperties.Mcp mcp = applicationProperties.getMcp();
+        List<String> allowed = mcp.getAllowedOperations();
+        List<String> blocked = mcp.getBlockedOperations();
+        if (blocked != null && blocked.contains(id)) {
+            return false;
+        }
+        if (allowed != null && !allowed.isEmpty()) {
+            return allowed.contains(id);
+        }
+        return true;
+    }
+
+    @EventListener(ContextRefreshedEvent.class)
+    public void discover() {
+        pdfOps.clear();
+        for (RequestMappingHandlerMapping mapping :
+                applicationContext.getBeansOfType(RequestMappingHandlerMapping.class).values()) {
+            for (Map.Entry<RequestMappingInfo, HandlerMethod> e :
+                    mapping.getHandlerMethods().entrySet()) {
+                indexOne(e.getKey(), e.getValue());
+            }
+        }
+        log.info("MCP tool catalog discovered {} PDF operation(s)", pdfOps.size());
+    }
+
+    private void indexOne(RequestMappingInfo info, HandlerMethod handler) {
+        Set<String> patterns = extractPatterns(info);
+        if (patterns.isEmpty()) {
+            return;
+        }
+        Set<RequestMethod> methods = info.getMethodsCondition().getMethods();
+        if (!isInvocableMethod(methods)) {
+            return;
+        }
+        for (String pattern : patterns) {
+            OperationCategory category = OperationCategory.fromUrl(pattern);
+            if (category == null) {
+                continue;
+            }
+            String opId = extractOpId(pattern, category);
+            if (opId == null) {
+                continue;
+            }
+            OperationMeta meta = buildMeta(opId, category, pattern, handler);
+            // First handler wins on duplicate URLs.
+            pdfOps.putIfAbsent(opId, meta);
+        }
+    }
+
+    private OperationMeta buildMeta(
+            String opId, OperationCategory category, String url, HandlerMethod handler) {
+        Method method = handler.getMethod();
+        Operation opAnno = method.getAnnotation(Operation.class);
+        String summary =
+                opAnno != null && !opAnno.summary().isBlank()
+                        ? opAnno.summary()
+                        : prettifyOpId(opId);
+        ObjectNode schema = paramSchemaFor(handler);
+        // Every mutating endpoint requires the write scope.
+        return new OperationMeta(
+                opId,
+                category,
+                summary,
+                schema,
+                WRITE_SCOPE,
+                OperationMeta.Target.JAVA_ENDPOINT,
+                url,
+                handler);
+    }
+
+    private ObjectNode paramSchemaFor(HandlerMethod handler) {
+        Optional<Class<?>> bodyType = firstComplexParamType(handler);
+        return bodyType.map(schemaGenerator::toSchema).orElseGet(() -> emptyObjectSchema());
+    }
+
+    private ObjectNode emptyObjectSchema() {
+        ObjectNode out = objectMapper.createObjectNode();
+        out.put("type", "object");
+        out.put("additionalProperties", true);
+        return out;
+    }
+
+    private Optional<Class<?>> firstComplexParamType(HandlerMethod handler) {
+        for (MethodParameter p : handler.getMethodParameters()) {
+            Class<?> type = p.getParameterType();
+            if (type.isPrimitive() || type == String.class || type.getName().startsWith("java.")) {
+                continue;
+            }
+            // Skip Spring-managed parameter types (HttpServletRequest, Principal, etc.).
+            String pkg = type.getPackageName();
+            if (pkg.startsWith("jakarta.") || pkg.startsWith("org.springframework.")) {
+                continue;
+            }
+            return Optional.of(type);
+        }
+        return Optional.empty();
+    }
+
+    public List<OperationMeta> enabledOps(OperationCategory category) {
+        if (category == OperationCategory.AI) {
+            List<OperationMeta> ai = new ArrayList<>();
+            for (OperationMeta m : aiOps.values()) {
+                if (isOperationAllowed(m.id())) {
+                    ai.add(m);
+                }
+            }
+            return ai;
+        }
+        List<OperationMeta> out = new ArrayList<>();
+        for (OperationMeta m : pdfOps.values()) {
+            if (m.category() == category
+                    && isOperationAllowed(m.id())
+                    && endpointConfiguration.isEndpointEnabledForUri(m.endpointPath())) {
+                out.add(m);
+            }
+        }
+        out.sort((a, b) -> a.id().compareTo(b.id()));
+        return out;
+    }
+
+    public Optional<OperationMeta> findByOperationId(String id) {
+        if (!isOperationAllowed(id)) {
+            return Optional.empty();
+        }
+        // A disabled PDF op returns empty rather than falling through to a same-id AI capability.
+        OperationMeta meta = pdfOps.get(id);
+        if (meta != null) {
+            boolean enabled =
+                    meta.target() != OperationMeta.Target.JAVA_ENDPOINT
+                            || endpointConfiguration.isEndpointEnabledForUri(meta.endpointPath());
+            return enabled ? Optional.of(meta) : Optional.empty();
+        }
+        return Optional.ofNullable(aiOps.get(id));
+    }
+
+    /** Replace the AI capabilities snapshot. Called by the engine refresh task. */
+    public void replaceAiCapabilities(Map<String, OperationMeta> updated) {
+        // Build a fresh map then swap atomically via the volatile reference. The previous
+        // implementation did putAll-then-retainAll on a shared ConcurrentHashMap, which left a
+        // transient window where readers could observe stale entries that should have been
+        // removed (race between the two structural updates).
+        Map<String, OperationMeta> next = new ConcurrentHashMap<>(updated);
+        this.aiOps = next;
+        log.info("MCP tool catalog AI capabilities replaced: {} entries", next.size());
+    }
+
+    /** Only POST/PUT endpoints are exposed as tools; DELETE and GET are excluded. */
+    static boolean isInvocableMethod(Set<RequestMethod> methods) {
+        return methods.contains(RequestMethod.POST) || methods.contains(RequestMethod.PUT);
+    }
+
+    private static String extractOpId(String pattern, OperationCategory category) {
+        if (category.urlPrefix() == null || !pattern.startsWith(category.urlPrefix())) {
+            return null;
+        }
+        String tail = pattern.substring(category.urlPrefix().length());
+        if (tail.isBlank() || tail.contains("/") || tail.contains("{")) {
+            // Skip nested paths and path-variable templates.
+            return null;
+        }
+        return tail;
+    }
+
+    private static String prettifyOpId(String id) {
+        return id.replace('-', ' ');
+    }
+
+    private static Set<String> extractPatterns(RequestMappingInfo info) {
+        try {
+            Method getDirectPaths = info.getClass().getMethod("getDirectPaths");
+            Object result = getDirectPaths.invoke(info);
+            if (result instanceof Set<?> set) {
+                Set<String> patterns = new TreeSet<>();
+                for (Object v : set) {
+                    if (v instanceof String s) {
+                        patterns.add(s);
+                    }
+                }
+                return patterns;
+            }
+        } catch (Exception e) {
+            log.trace("getDirectPaths unavailable on RequestMappingInfo", e);
+        }
+        return Collections.emptySet();
+    }
+
+    public Map<String, OperationMeta> snapshotPdfOps() {
+        return new LinkedHashMap<>(pdfOps);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/catalog/OperationCategory.java

@@ -0,0 +1,38 @@
+package stirling.software.proprietary.mcp.catalog;
+
+/** MCP tool categories; {@link #urlPrefix} maps a {@code /api/v1/} namespace to a category. */
+public enum OperationCategory {
+    CONVERT("/api/v1/convert/", "stirling_convert"),
+    PAGES("/api/v1/general/", "stirling_pages"),
+    MISC("/api/v1/misc/", "stirling_misc"),
+    SECURITY("/api/v1/security/", "stirling_security"),
+    AI(null, "stirling_ai");
+
+    private final String urlPrefix;
+    private final String toolName;
+
+    OperationCategory(String urlPrefix, String toolName) {
+        this.urlPrefix = urlPrefix;
+        this.toolName = toolName;
+    }
+
+    public String urlPrefix() {
+        return urlPrefix;
+    }
+
+    public String toolName() {
+        return toolName;
+    }
+
+    public static OperationCategory fromUrl(String url) {
+        if (url == null) {
+            return null;
+        }
+        for (OperationCategory c : values()) {
+            if (c.urlPrefix != null && url.startsWith(c.urlPrefix)) {
+                return c;
+            }
+        }
+        return null;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/catalog/OperationMeta.java

@@ -0,0 +1,22 @@
+package stirling.software.proprietary.mcp.catalog;
+
+import org.springframework.web.method.HandlerMethod;
+
+import tools.jackson.databind.node.ObjectNode;
+
+/** Metadata for one MCP-exposed operation (PDF endpoint or AI capability). */
+public record OperationMeta(
+        String id,
+        OperationCategory category,
+        String summary,
+        ObjectNode paramSchema,
+        String requiredScope,
+        Target target,
+        String endpointPath,
+        HandlerMethod handlerMethod) {
+
+    public enum Target {
+        JAVA_ENDPOINT,
+        ENGINE_CAPABILITY
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/catalog/SimpleSchemaGenerator.java

@@ -0,0 +1,178 @@
+package stirling.software.proprietary.mcp.catalog;
+
+import java.lang.reflect.Field;
+import java.lang.reflect.ParameterizedType;
+import java.lang.reflect.Type;
+import java.util.ArrayList;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+import org.springframework.web.multipart.MultipartFile;
+
+import com.fasterxml.jackson.annotation.JsonIgnore;
+import com.fasterxml.jackson.annotation.JsonProperty;
+
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ArrayNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Reflection-based JSON Schema generator for controller request-body classes. {@link MultipartFile}
+ * fields are emitted as {@code "type":"string"} with a {@code "format":"file-id"} hint.
+ */
+public final class SimpleSchemaGenerator {
+
+    private final ObjectMapper mapper;
+
+    public SimpleSchemaGenerator(ObjectMapper mapper) {
+        this.mapper = mapper;
+    }
+
+    public ObjectNode toSchema(Class<?> type) {
+        return toSchema(type, new HashSet<>());
+    }
+
+    private ObjectNode toSchema(Class<?> type, Set<Class<?>> visited) {
+        ObjectNode schema = mapper.createObjectNode();
+        schema.put("type", "object");
+        schema.put("additionalProperties", false);
+        ObjectNode properties = schema.putObject("properties");
+        ArrayNode required = mapper.createArrayNode();
+        if (!visited.add(type)) {
+            // Cycle: emit a loose object and bail.
+            schema.put("additionalProperties", true);
+            return schema;
+        }
+
+        Set<String> seen = new HashSet<>();
+        for (Field field : collectFields(type)) {
+            if (java.lang.reflect.Modifier.isStatic(field.getModifiers())
+                    || java.lang.reflect.Modifier.isTransient(field.getModifiers())) {
+                continue;
+            }
+            // Skip fields Jackson won't (de)serialize.
+            if (field.isAnnotationPresent(JsonIgnore.class)) {
+                continue;
+            }
+            String name = jsonPropertyName(field);
+            if (!seen.add(name)) {
+                continue;
+            }
+            properties.set(name, typeSchema(field.getGenericType(), visited));
+            if (isRequired(field)) {
+                required.add(name);
+            }
+        }
+
+        if (!required.isEmpty()) {
+            schema.set("required", required);
+        }
+        return schema;
+    }
+
+    private List<Field> collectFields(Class<?> type) {
+        List<Field> all = new ArrayList<>();
+        for (Class<?> c = type; c != null && c != Object.class; c = c.getSuperclass()) {
+            for (Field f : c.getDeclaredFields()) {
+                all.add(f);
+            }
+        }
+        return all;
+    }
+
+    private static String jsonPropertyName(Field field) {
+        JsonProperty ann = field.getAnnotation(JsonProperty.class);
+        if (ann != null && !ann.value().isEmpty()) {
+            return ann.value();
+        }
+        return field.getName();
+    }
+
+    private boolean isRequired(Field field) {
+        JsonProperty json = field.getAnnotation(JsonProperty.class);
+        if (json != null && json.required()) {
+            return true;
+        }
+        return field.isAnnotationPresent(jakarta.validation.constraints.NotNull.class)
+                || field.isAnnotationPresent(jakarta.validation.constraints.NotBlank.class)
+                || field.isAnnotationPresent(jakarta.validation.constraints.NotEmpty.class);
+    }
+
+    private ObjectNode typeSchema(Type t, Set<Class<?>> visited) {
+        ObjectNode out = mapper.createObjectNode();
+        if (t instanceof Class<?> c) {
+            populatePrimitive(out, c, visited);
+        } else if (t instanceof ParameterizedType pt) {
+            Type raw = pt.getRawType();
+            if (raw instanceof Class<?> rawClass) {
+                if (java.util.Collection.class.isAssignableFrom(rawClass)) {
+                    out.put("type", "array");
+                    Type[] args = pt.getActualTypeArguments();
+                    if (args.length == 1) {
+                        out.set("items", typeSchema(args[0], visited));
+                    }
+                } else if (java.util.Map.class.isAssignableFrom(rawClass)) {
+                    out.put("type", "object");
+                    out.put("additionalProperties", true);
+                } else {
+                    populatePrimitive(out, rawClass, visited);
+                }
+            } else {
+                out.put("type", "object");
+            }
+        } else {
+            out.put("type", "object");
+        }
+        return out;
+    }
+
+    private void populatePrimitive(ObjectNode out, Class<?> c, Set<Class<?>> visited) {
+        if (MultipartFile.class.isAssignableFrom(c)) {
+            out.put("type", "string");
+            out.put("format", "file-id");
+            out.put(
+                    "description",
+                    "Reference to a previously-uploaded file in Stirling's job store.");
+            return;
+        }
+        if (c.isArray()) {
+            out.put("type", "array");
+            out.set("items", typeSchema(c.getComponentType(), visited));
+            return;
+        }
+        if (c == String.class) {
+            out.put("type", "string");
+        } else if (c == boolean.class || c == Boolean.class) {
+            out.put("type", "boolean");
+        } else if (c == int.class
+                || c == Integer.class
+                || c == long.class
+                || c == Long.class
+                || c == short.class
+                || c == Short.class
+                || c == byte.class
+                || c == Byte.class) {
+            out.put("type", "integer");
+        } else if (c == float.class || c == Float.class || c == double.class || c == Double.class) {
+            out.put("type", "number");
+        } else if (c.isEnum()) {
+            out.put("type", "string");
+            ArrayNode values = out.putArray("enum");
+            for (Object constant : c.getEnumConstants()) {
+                values.add(constant.toString());
+            }
+        } else if (c == java.util.UUID.class) {
+            out.put("type", "string");
+            out.put("format", "uuid");
+        } else if (java.time.temporal.Temporal.class.isAssignableFrom(c)
+                || c == java.util.Date.class) {
+            out.put("type", "string");
+            out.put("format", "date-time");
+        } else {
+            // Complex bean: recurse with the shared visited set.
+            ObjectNode nested = toSchema(c, visited);
+            out.setAll(nested);
+        }
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/engine/EngineCapabilityClient.java

@@ -0,0 +1,204 @@
+package stirling.software.proprietary.mcp.engine;
+
+import java.io.IOException;
+import java.net.URI;
+import java.net.http.HttpClient;
+import java.net.http.HttpRequest;
+import java.net.http.HttpResponse;
+import java.time.Duration;
+import java.util.LinkedHashMap;
+import java.util.Map;
+import java.util.concurrent.Executors;
+import java.util.concurrent.ScheduledExecutorService;
+import java.util.concurrent.TimeUnit;
+
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.boot.context.event.ApplicationReadyEvent;
+import org.springframework.context.event.EventListener;
+import org.springframework.stereotype.Component;
+
+import jakarta.annotation.PostConstruct;
+import jakarta.annotation.PreDestroy;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationCategory;
+import stirling.software.proprietary.mcp.catalog.OperationMeta;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Pulls the engine's capabilities manifest at boot and on a schedule, feeding it into the shared
+ * {@link McpToolCatalog}.
+ */
+@Slf4j
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class EngineCapabilityClient {
+
+    private final ApplicationProperties applicationProperties;
+    private final McpToolCatalog catalog;
+    private final ObjectMapper mapper;
+    private final HttpClient httpClient;
+    private final String sharedSecret;
+
+    private ScheduledExecutorService scheduler;
+
+    public EngineCapabilityClient(
+            ApplicationProperties applicationProperties,
+            McpToolCatalog catalog,
+            ObjectMapper mapper) {
+        this.applicationProperties = applicationProperties;
+        this.catalog = catalog;
+        this.mapper = mapper;
+        this.httpClient = HttpClient.newBuilder().connectTimeout(Duration.ofSeconds(5)).build();
+        this.sharedSecret = System.getenv("STIRLING_ENGINE_SHARED_SECRET");
+    }
+
+    @PostConstruct
+    void start() {
+        scheduler =
+                Executors.newSingleThreadScheduledExecutor(
+                        r -> {
+                            Thread t = new Thread(r, "mcp-engine-capability-refresh");
+                            t.setDaemon(true);
+                            return t;
+                        });
+    }
+
+    @EventListener(ApplicationReadyEvent.class)
+    public void onReady() {
+        long minutes =
+                Math.max(1, applicationProperties.getMcp().getEngineCapabilityRefreshMinutes());
+        // First refresh immediately, then on the configured cadence.
+        scheduler.schedule(this::refreshSafely, 0, TimeUnit.SECONDS);
+        scheduler.scheduleAtFixedRate(this::refreshSafely, minutes, minutes, TimeUnit.MINUTES);
+        log.info("MCP engine capability refresh scheduled every {} minute(s)", minutes);
+    }
+
+    @PreDestroy
+    void stop() {
+        if (scheduler != null) {
+            scheduler.shutdownNow();
+        }
+    }
+
+    private void refreshSafely() {
+        try {
+            refresh();
+        } catch (Exception e) {
+            log.warn(
+                    "MCP engine capability refresh failed ({}). AI tool enum stays at the last"
+                            + " known state until the next successful pull.",
+                    e.getMessage());
+        }
+    }
+
+    /** Visible for testing. */
+    public void refresh() throws IOException, InterruptedException {
+        if (!applicationProperties.getAiEngine().isEnabled()) {
+            log.debug("AI engine disabled; skipping MCP capability refresh");
+            catalog.replaceAiCapabilities(Map.of());
+            return;
+        }
+        // Trim whitespace and any trailing slash to avoid a malformed URI.
+        String base = applicationProperties.getAiEngine().getUrl().strip().replaceAll("/+$", "");
+        URI uri = URI.create(base + "/api/v1/agents/capabilities");
+        HttpRequest.Builder reqBuilder =
+                HttpRequest.newBuilder()
+                        .uri(uri)
+                        .timeout(Duration.ofSeconds(10))
+                        .header("Accept", "application/json")
+                        .GET();
+        if (sharedSecret != null && !sharedSecret.isBlank()) {
+            reqBuilder.header("X-Engine-Auth", sharedSecret);
+        }
+        HttpResponse<String> response =
+                httpClient.send(reqBuilder.build(), HttpResponse.BodyHandlers.ofString());
+        if (response.statusCode() != 200) {
+            throw new IOException(
+                    "Engine capabilities endpoint returned HTTP " + response.statusCode());
+        }
+        Map<String, OperationMeta> parsed = parseManifest(response.body());
+        catalog.replaceAiCapabilities(parsed);
+    }
+
+    private Map<String, OperationMeta> parseManifest(String body) throws IOException {
+        JsonNode root = mapper.readTree(body);
+        JsonNode capabilities = root.get("capabilities");
+        if (capabilities == null || !capabilities.isArray()) {
+            throw new IOException("Manifest missing 'capabilities' array");
+        }
+        Map<String, OperationMeta> out = new LinkedHashMap<>();
+        for (JsonNode entry : capabilities) {
+            JsonNode id = entry.get("id");
+            JsonNode desc = entry.get("description");
+            JsonNode schema = entry.get("input_schema");
+            JsonNode scope = entry.get("required_scope");
+            JsonNode route = entry.get("route");
+            if (id == null || !id.isTextual() || schema == null || !schema.isObject()) {
+                log.warn("Skipping malformed capability entry: {}", entry);
+                continue;
+            }
+            String routeValue = route == null || !route.isTextual() ? null : route.asText();
+            if (routeValue != null && !isSafeRelativeRoute(routeValue)) {
+                // Defence in depth: a tampered manifest must not steer Java at an arbitrary
+                // host/path.
+                log.warn(
+                        "Skipping capability '{}' with unsafe route '{}' (must be a server-relative"
+                                + " /api path with no scheme, authority, or '..')",
+                        id.asText(),
+                        routeValue);
+                continue;
+            }
+            // Fail safe: default to the stricter write scope when the manifest omits one.
+            String requiredScope =
+                    scope != null && scope.isTextual() && !scope.asText().isBlank()
+                            ? scope.asText()
+                            : WRITE_SCOPE;
+            ObjectNode schemaCopy = (ObjectNode) schema.deepCopy();
+            out.put(
+                    id.asText(),
+                    new OperationMeta(
+                            id.asText(),
+                            OperationCategory.AI,
+                            desc == null ? id.asText() : desc.asText(),
+                            schemaCopy,
+                            requiredScope,
+                            OperationMeta.Target.ENGINE_CAPABILITY,
+                            routeValue,
+                            null));
+        }
+        return out;
+    }
+
+    private static final String WRITE_SCOPE = "mcp.tools.write";
+
+    /**
+     * True only for a server-relative {@code /api/} path with no scheme, authority, {@code ..}, or
+     * control chars (blocks SSRF / path escape).
+     */
+    static boolean isSafeRelativeRoute(String route) {
+        if (route == null || route.isBlank() || !route.startsWith("/api/")) {
+            return false;
+        }
+        if (route.startsWith("//")
+                || route.contains("..")
+                || route.contains("@")
+                || route.contains("\\")
+                || route.contains(":")) {
+            return false;
+        }
+        for (int i = 0; i < route.length(); i++) {
+            char c = route.charAt(i);
+            if (c <= ' ') {
+                return false;
+            }
+        }
+        return true;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/jsonrpc/JsonRpcError.java

@@ -0,0 +1,36 @@
+package stirling.software.proprietary.mcp.jsonrpc;
+
+import com.fasterxml.jackson.annotation.JsonInclude;
+
+import tools.jackson.databind.JsonNode;
+
+/** JSON-RPC 2.0 error object. */
+@JsonInclude(JsonInclude.Include.NON_NULL)
+public record JsonRpcError(int code, String message, JsonNode data) {
+
+    public static final int PARSE_ERROR = -32700;
+    public static final int INVALID_REQUEST = -32600;
+    public static final int METHOD_NOT_FOUND = -32601;
+    public static final int INVALID_PARAMS = -32602;
+    public static final int INTERNAL_ERROR = -32603;
+
+    public static JsonRpcError parseError(String message) {
+        return new JsonRpcError(PARSE_ERROR, message, null);
+    }
+
+    public static JsonRpcError invalidRequest(String message) {
+        return new JsonRpcError(INVALID_REQUEST, message, null);
+    }
+
+    public static JsonRpcError methodNotFound(String method) {
+        return new JsonRpcError(METHOD_NOT_FOUND, "Method not found: " + method, null);
+    }
+
+    public static JsonRpcError invalidParams(String message) {
+        return new JsonRpcError(INVALID_PARAMS, message, null);
+    }
+
+    public static JsonRpcError internalError(String message) {
+        return new JsonRpcError(INTERNAL_ERROR, message, null);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/jsonrpc/JsonRpcRequest.java

@@ -0,0 +1,14 @@
+package stirling.software.proprietary.mcp.jsonrpc;
+
+import com.fasterxml.jackson.annotation.JsonInclude;
+
+import tools.jackson.databind.JsonNode;
+
+/** JSON-RPC 2.0 request frame; a null {@code id} marks a notification. */
+@JsonInclude(JsonInclude.Include.NON_NULL)
+public record JsonRpcRequest(String jsonrpc, JsonNode id, String method, JsonNode params) {
+
+    public boolean isNotification() {
+        return id == null || id.isNull();
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/jsonrpc/JsonRpcResponse.java

@@ -0,0 +1,18 @@
+package stirling.software.proprietary.mcp.jsonrpc;
+
+import com.fasterxml.jackson.annotation.JsonInclude;
+
+import tools.jackson.databind.JsonNode;
+
+/** JSON-RPC 2.0 response; exactly one of {@code result} or {@code error} is non-null. */
+@JsonInclude(JsonInclude.Include.NON_NULL)
+public record JsonRpcResponse(String jsonrpc, JsonNode id, Object result, JsonRpcError error) {
+
+    public static JsonRpcResponse success(JsonNode id, Object result) {
+        return new JsonRpcResponse("2.0", id, result, null);
+    }
+
+    public static JsonRpcResponse failure(JsonNode id, JsonRpcError error) {
+        return new JsonRpcResponse("2.0", id, null, error);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/security/McpApiKeyAuthFilter.java

@@ -0,0 +1,85 @@
+package stirling.software.proprietary.mcp.security;
+
+import java.io.IOException;
+import java.util.List;
+import java.util.Optional;
+
+import org.springframework.security.authentication.AnonymousAuthenticationToken;
+import org.springframework.security.authentication.UsernamePasswordAuthenticationToken;
+import org.springframework.security.core.Authentication;
+import org.springframework.security.core.GrantedAuthority;
+import org.springframework.security.core.authority.SimpleGrantedAuthority;
+import org.springframework.security.core.context.SecurityContext;
+import org.springframework.security.core.context.SecurityContextHolder;
+import org.springframework.web.filter.OncePerRequestFilter;
+
+import jakarta.servlet.FilterChain;
+import jakarta.servlet.ServletException;
+import jakarta.servlet.http.HttpServletRequest;
+import jakarta.servlet.http.HttpServletResponse;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.proprietary.security.model.User;
+import stirling.software.proprietary.security.service.UserService;
+
+/**
+ * API-key auth for the MCP endpoint: validates a Stirling per-user API key and binds the request to
+ * that user with the MCP scopes.
+ */
+@Slf4j
+public class McpApiKeyAuthFilter extends OncePerRequestFilter {
+
+    private static final List<GrantedAuthority> MCP_SCOPES =
+            List.of(
+                    new SimpleGrantedAuthority("SCOPE_mcp.tools.read"),
+                    new SimpleGrantedAuthority("SCOPE_mcp.tools.write"));
+
+    private final UserService userService;
+
+    public McpApiKeyAuthFilter(UserService userService) {
+        this.userService = userService;
+    }
+
+    @Override
+    protected void doFilterInternal(
+            HttpServletRequest request, HttpServletResponse response, FilterChain filterChain)
+            throws ServletException, IOException {
+        Authentication existing = SecurityContextHolder.getContext().getAuthentication();
+        // Treat an anonymous token as not authenticated so the key is still processed.
+        boolean unauthenticated =
+                existing == null
+                        || existing instanceof AnonymousAuthenticationToken
+                        || !existing.isAuthenticated();
+        if (unauthenticated) {
+            String apiKey = extractKey(request);
+            if (apiKey != null && !apiKey.isBlank()) {
+                Optional<User> user = userService.getUserByApiKey(apiKey);
+                if (user.isPresent() && user.get().isEnabled()) {
+                    UsernamePasswordAuthenticationToken auth =
+                            new UsernamePasswordAuthenticationToken(
+                                    user.get().getUsername(), null, MCP_SCOPES);
+                    SecurityContext context = SecurityContextHolder.createEmptyContext();
+                    context.setAuthentication(auth);
+                    SecurityContextHolder.setContext(context);
+                } else {
+                    log.warn(
+                            "MCP access denied: presented API key did not match an active account");
+                }
+            }
+        }
+        filterChain.doFilter(request, response);
+    }
+
+    private String extractKey(HttpServletRequest request) {
+        String headerKey = request.getHeader("X-API-KEY");
+        if (headerKey != null && !headerKey.isBlank()) {
+            return headerKey.trim();
+        }
+        String authz = request.getHeader("Authorization");
+        if (authz != null && authz.regionMatches(true, 0, "Bearer ", 0, 7)) {
+            return authz.substring(7).trim();
+        }
+        return null;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/security/McpAudienceValidator.java

@@ -0,0 +1,44 @@
+package stirling.software.proprietary.mcp.security;
+
+import java.util.List;
+
+import org.springframework.security.oauth2.core.OAuth2Error;
+import org.springframework.security.oauth2.core.OAuth2TokenValidator;
+import org.springframework.security.oauth2.core.OAuth2TokenValidatorResult;
+import org.springframework.security.oauth2.jwt.Jwt;
+
+/**
+ * RFC 8707 audience binding: a JWT at the MCP endpoint must list this server's resource id in its
+ * {@code aud} claim. Fails closed when the resource id is unset.
+ */
+public class McpAudienceValidator implements OAuth2TokenValidator<Jwt> {
+
+    private final String expectedResourceId;
+
+    public McpAudienceValidator(String expectedResourceId) {
+        this.expectedResourceId = expectedResourceId == null ? "" : expectedResourceId;
+    }
+
+    @Override
+    public OAuth2TokenValidatorResult validate(Jwt token) {
+        if (expectedResourceId.isBlank()) {
+            return OAuth2TokenValidatorResult.failure(
+                    new OAuth2Error(
+                            "invalid_token",
+                            "MCP server has no resource id configured; rejecting all tokens"
+                                    + " until mcp.auth.resource-id is set.",
+                            null));
+        }
+        List<String> aud = token.getAudience();
+        if (aud == null || !aud.contains(expectedResourceId)) {
+            return OAuth2TokenValidatorResult.failure(
+                    new OAuth2Error(
+                            "invalid_token",
+                            "Token audience does not include this server's resource id ("
+                                    + expectedResourceId
+                                    + ").",
+                            null));
+        }
+        return OAuth2TokenValidatorResult.success();
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/security/McpAuthenticationEntryPoint.java

@@ -0,0 +1,76 @@
+package stirling.software.proprietary.mcp.security;
+
+import java.io.IOException;
+
+import org.springframework.http.HttpStatus;
+import org.springframework.security.core.AuthenticationException;
+import org.springframework.security.web.AuthenticationEntryPoint;
+
+import jakarta.servlet.http.HttpServletRequest;
+import jakarta.servlet.http.HttpServletResponse;
+
+/**
+ * Emits 401 + {@code WWW-Authenticate: Bearer resource_metadata="..."} (RFC 9728), preferring
+ * X-Forwarded-* headers to build the public-facing metadata URL.
+ */
+public class McpAuthenticationEntryPoint implements AuthenticationEntryPoint {
+
+    private final String metadataPath;
+
+    public McpAuthenticationEntryPoint(String metadataPath) {
+        this.metadataPath =
+                metadataPath == null ? "/.well-known/oauth-protected-resource" : metadataPath;
+    }
+
+    @Override
+    public void commence(
+            HttpServletRequest request,
+            HttpServletResponse response,
+            AuthenticationException authException)
+            throws IOException {
+        String scheme = firstForwarded(request, "X-Forwarded-Proto", request.getScheme());
+        String authority = forwardedHost(request, scheme);
+        String metadataUrl = scheme + "://" + authority + metadataPath;
+        response.setHeader(
+                "WWW-Authenticate",
+                "Bearer error=\"invalid_token\", resource_metadata=\"" + metadataUrl + "\"");
+        response.sendError(HttpStatus.UNAUTHORIZED.value(), "Unauthorized");
+    }
+
+    /** host[:port] from forwarded headers when present, else the servlet host/port. */
+    private static String forwardedHost(HttpServletRequest request, String scheme) {
+        String host = firstForwarded(request, "X-Forwarded-Host", null);
+        if (host != null && !host.isBlank()) {
+            // X-Forwarded-Host may already carry a port.
+            if (host.contains(":")) {
+                return host;
+            }
+            String fwdPort = firstForwarded(request, "X-Forwarded-Port", null);
+            if (fwdPort != null && !isDefaultPort(scheme, fwdPort)) {
+                return host + ":" + fwdPort;
+            }
+            return host;
+        }
+        String authority = request.getServerName();
+        int port = request.getServerPort();
+        if (port > 0 && !isDefaultPort(scheme, Integer.toString(port))) {
+            authority = authority + ":" + port;
+        }
+        return authority;
+    }
+
+    /** First (client-most) value of a possibly comma-listed forwarded header, trimmed. */
+    private static String firstForwarded(HttpServletRequest request, String name, String fallback) {
+        String value = request.getHeader(name);
+        if (value == null || value.isBlank()) {
+            return fallback;
+        }
+        int comma = value.indexOf(',');
+        return (comma >= 0 ? value.substring(0, comma) : value).trim();
+    }
+
+    private static boolean isDefaultPort(String scheme, String port) {
+        return ("http".equals(scheme) && "80".equals(port))
+                || ("https".equals(scheme) && "443".equals(port));
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/security/McpRequestSizeFilter.java

@@ -0,0 +1,128 @@
+package stirling.software.proprietary.mcp.security;
+
+import java.io.BufferedReader;
+import java.io.ByteArrayInputStream;
+import java.io.ByteArrayOutputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.InputStreamReader;
+import java.nio.charset.Charset;
+import java.nio.charset.StandardCharsets;
+
+import org.springframework.web.filter.OncePerRequestFilter;
+
+import jakarta.servlet.FilterChain;
+import jakarta.servlet.ReadListener;
+import jakarta.servlet.ServletException;
+import jakarta.servlet.ServletInputStream;
+import jakarta.servlet.http.HttpServletRequest;
+import jakarta.servlet.http.HttpServletRequestWrapper;
+import jakarta.servlet.http.HttpServletResponse;
+
+/**
+ * Caps MCP request body size (via Content-Length and by buffering up to the cap) and rejects
+ * oversized bodies with a clean 413 before JSON parsing.
+ */
+public class McpRequestSizeFilter extends OncePerRequestFilter {
+
+    private final long maxBodyBytes;
+
+    public McpRequestSizeFilter(long maxBodyBytes) {
+        this.maxBodyBytes = maxBodyBytes > 0 ? maxBodyBytes : 256L * 1024L;
+    }
+
+    @Override
+    protected void doFilterInternal(
+            HttpServletRequest request, HttpServletResponse response, FilterChain filterChain)
+            throws ServletException, IOException {
+        long declared = request.getContentLengthLong();
+        if (declared > maxBodyBytes) {
+            tooLarge(response);
+            return;
+        }
+        byte[] body;
+        try {
+            body = readUpTo(request.getInputStream(), maxBodyBytes);
+        } catch (BodyTooLargeException e) {
+            tooLarge(response);
+            return;
+        }
+        filterChain.doFilter(new CachedBodyRequest(request, body), response);
+    }
+
+    private static byte[] readUpTo(InputStream in, long max) throws IOException {
+        ByteArrayOutputStream buffer = new ByteArrayOutputStream();
+        byte[] chunk = new byte[8192];
+        long total = 0;
+        int n;
+        while ((n = in.read(chunk)) != -1) {
+            total += n;
+            if (total > max) {
+                throw new BodyTooLargeException();
+            }
+            buffer.write(chunk, 0, n);
+        }
+        return buffer.toByteArray();
+    }
+
+    private void tooLarge(HttpServletResponse response) throws IOException {
+        response.setStatus(HttpServletResponse.SC_REQUEST_ENTITY_TOO_LARGE);
+        response.setContentType("application/json");
+        response.getWriter()
+                .write(
+                        "{\"error\":\"payload_too_large\",\"message\":\"MCP request body exceeds the"
+                                + " configured limit of "
+                                + maxBodyBytes
+                                + " bytes.\"}");
+    }
+
+    private static final class BodyTooLargeException extends IOException {}
+
+    /** Re-serves the buffered body to the controller. */
+    private static final class CachedBodyRequest extends HttpServletRequestWrapper {
+        private final byte[] body;
+
+        CachedBodyRequest(HttpServletRequest request, byte[] body) {
+            super(request);
+            this.body = body;
+        }
+
+        @Override
+        public ServletInputStream getInputStream() {
+            ByteArrayInputStream source = new ByteArrayInputStream(body);
+            return new ServletInputStream() {
+                @Override
+                public int read() {
+                    return source.read();
+                }
+
+                @Override
+                public int read(byte[] b, int off, int len) {
+                    return source.read(b, off, len);
+                }
+
+                @Override
+                public boolean isFinished() {
+                    return source.available() == 0;
+                }
+
+                @Override
+                public boolean isReady() {
+                    return true;
+                }
+
+                @Override
+                public void setReadListener(ReadListener readListener) {
+                    // Synchronous buffered body; no async reads.
+                }
+            };
+        }
+
+        @Override
+        public BufferedReader getReader() {
+            String enc = getCharacterEncoding();
+            Charset cs = enc == null ? StandardCharsets.UTF_8 : Charset.forName(enc);
+            return new BufferedReader(new InputStreamReader(new ByteArrayInputStream(body), cs));
+        }
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/security/McpSecurityConfig.java

@@ -0,0 +1,262 @@
+package stirling.software.proprietary.mcp.security;
+
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.List;
+
+import org.springframework.beans.factory.ObjectProvider;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.context.annotation.Bean;
+import org.springframework.context.annotation.Configuration;
+import org.springframework.context.annotation.Lazy;
+import org.springframework.core.Ordered;
+import org.springframework.core.annotation.Order;
+import org.springframework.core.convert.converter.Converter;
+import org.springframework.http.HttpMethod;
+import org.springframework.security.authentication.AbstractAuthenticationToken;
+import org.springframework.security.config.annotation.web.builders.HttpSecurity;
+import org.springframework.security.config.http.SessionCreationPolicy;
+import org.springframework.security.core.GrantedAuthority;
+import org.springframework.security.core.authority.SimpleGrantedAuthority;
+import org.springframework.security.oauth2.core.DelegatingOAuth2TokenValidator;
+import org.springframework.security.oauth2.core.OAuth2TokenValidator;
+import org.springframework.security.oauth2.jwt.Jwt;
+import org.springframework.security.oauth2.jwt.JwtDecoder;
+import org.springframework.security.oauth2.jwt.JwtValidators;
+import org.springframework.security.oauth2.jwt.NimbusJwtDecoder;
+import org.springframework.security.oauth2.server.resource.authentication.JwtAuthenticationConverter;
+import org.springframework.security.oauth2.server.resource.authentication.JwtGrantedAuthoritiesConverter;
+import org.springframework.security.oauth2.server.resource.web.authentication.BearerTokenAuthenticationFilter;
+import org.springframework.security.web.SecurityFilterChain;
+import org.springframework.security.web.access.intercept.AuthorizationFilter;
+import org.springframework.security.web.authentication.AnonymousAuthenticationFilter;
+import org.springframework.web.cors.CorsConfigurationSource;
+
+import jakarta.annotation.PostConstruct;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.proprietary.security.service.UserService;
+
+/**
+ * MCP security chain: validates JWTs (JWKS + RFC 8707 audience), maps scope claims to authorities,
+ * and fails closed when the issuer is unset.
+ */
+@Slf4j
+@Configuration
+@Order(Ordered.HIGHEST_PRECEDENCE)
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class McpSecurityConfig {
+
+    private final ApplicationProperties applicationProperties;
+    private final UserService userService;
+
+    // Reuse the app's CORS config; ObjectProvider so the chain still wires when no CORS bean
+    // exists.
+    private final ObjectProvider<CorsConfigurationSource> corsConfigurationSource;
+
+    private static final String BASE_PATH = "/mcp";
+
+    public McpSecurityConfig(
+            ApplicationProperties applicationProperties,
+            @Lazy UserService userService,
+            ObjectProvider<CorsConfigurationSource> corsConfigurationSource) {
+        this.applicationProperties = applicationProperties;
+        this.userService = userService;
+        this.corsConfigurationSource = corsConfigurationSource;
+    }
+
+    /** Enable CORS on the MCP chain using the app-wide source when available. */
+    private void applyCors(HttpSecurity http) throws Exception {
+        CorsConfigurationSource source = corsConfigurationSource.getIfAvailable();
+        if (source != null) {
+            http.cors(cors -> cors.configurationSource(source));
+        }
+    }
+
+    @PostConstruct
+    void warnIfMisconfigured() {
+        ApplicationProperties.Mcp mcp = applicationProperties.getMcp();
+        if (isApiKeyMode()) {
+            log.info(
+                    "MCP auth mode = apikey: clients authenticate with a Stirling per-user API key"
+                            + " (X-API-KEY header). No OAuth issuer required.");
+        } else {
+            if (mcp.getAuth().getIssuerUri().isBlank()) {
+                log.warn(
+                        "MCP enabled but mcp.auth.issuer-uri is blank - JWT decoder will reject"
+                                + " every token (fail-closed). Set mcp.auth.issuer-uri and"
+                                + " mcp.auth.resource-id before exposing /mcp to clients.");
+            }
+            if (mcp.getAuth().getResourceId().isBlank()) {
+                log.warn(
+                        "MCP enabled but mcp.auth.resource-id is blank - audience validator will"
+                                + " reject every token. Set this to the public URL of the MCP"
+                                + " endpoint (RFC 8707).");
+            }
+        }
+    }
+
+    @Bean
+    @Order(0)
+    SecurityFilterChain mcpSecurityFilterChain(HttpSecurity http, JwtDecoder mcpJwtDecoder)
+            throws Exception {
+        ApplicationProperties.Mcp.Auth auth = applicationProperties.getMcp().getAuth();
+        if (isApiKeyMode()) {
+            return apiKeyFilterChain(http);
+        }
+        return oauthFilterChain(http, mcpJwtDecoder, auth);
+    }
+
+    private boolean isApiKeyMode() {
+        return "apikey".equalsIgnoreCase(applicationProperties.getMcp().getAuth().getMode());
+    }
+
+    /**
+     * API-key chain: a Stirling per-user API key is validated by {@link McpApiKeyAuthFilter};
+     * otherwise 401.
+     */
+    private SecurityFilterChain apiKeyFilterChain(HttpSecurity http) throws Exception {
+        applyCors(http);
+        http.securityMatcher(BASE_PATH, BASE_PATH + "/**")
+                // CSRF intentionally disabled: /mcp is a stateless JSON-RPC API authenticated by an
+                // out-of-band X-API-KEY header (or Authorization: Bearer <key>). No cookies, no
+                // session, no form submissions; a browser cannot trick a victim into sending the
+                // header cross-origin, so the CSRF attack model does not apply. CodeQL flags this
+                // generically; the SessionCreationPolicy.STATELESS below is the relevant guarantee.
+                .csrf(csrf -> csrf.disable())
+                .sessionManagement(s -> s.sessionCreationPolicy(SessionCreationPolicy.STATELESS))
+                .authorizeHttpRequests(a -> a.anyRequest().authenticated())
+                .exceptionHandling(
+                        e ->
+                                e.authenticationEntryPoint(
+                                        (request, response, ex) -> {
+                                            response.setStatus(401);
+                                            response.setHeader(
+                                                    "WWW-Authenticate",
+                                                    "Bearer realm=\"Stirling MCP (API key)\"");
+                                            response.setContentType("application/json");
+                                            response.getWriter()
+                                                    .write(
+                                                            "{\"error\":\"unauthorized\",\"message\":\"Provide a valid Stirling API key via the X-API-KEY header (or Authorization: Bearer <key>).\"}");
+                                        }))
+                .addFilterBefore(
+                        new McpRequestSizeFilter(
+                                applicationProperties.getMcp().getMaxRequestBytes()),
+                        AuthorizationFilter.class)
+                // Authenticate before the anonymous filter sets an anonymous token.
+                .addFilterBefore(
+                        new McpApiKeyAuthFilter(userService), AnonymousAuthenticationFilter.class);
+        return http.build();
+    }
+
+    /** OAuth2 resource-server chain (JWT, RFC 8707 audience, RFC 9728 metadata). */
+    private SecurityFilterChain oauthFilterChain(
+            HttpSecurity http, JwtDecoder mcpJwtDecoder, ApplicationProperties.Mcp.Auth auth)
+            throws Exception {
+        String metadataPath = "/.well-known/oauth-protected-resource";
+        applyCors(http);
+        http.securityMatcher(BASE_PATH, BASE_PATH + "/**", metadataPath)
+                // CSRF intentionally disabled: /mcp is a stateless JSON-RPC resource server
+                // authenticated by OAuth2 Bearer JWTs (Authorization header). No cookies, no
+                // session, no form submissions; CSRF requires browser-attached ambient credentials
+                // and the bearer token is supplied per-request by the MCP client. CodeQL flags
+                // this generically; the SessionCreationPolicy.STATELESS below is the actual
+                // guarantee, and the .well-known metadata endpoint only serves GET.
+                .csrf(csrf -> csrf.disable())
+                .sessionManagement(s -> s.sessionCreationPolicy(SessionCreationPolicy.STATELESS))
+                .authorizeHttpRequests(
+                        a ->
+                                a.requestMatchers(HttpMethod.GET, metadataPath)
+                                        .permitAll()
+                                        .anyRequest()
+                                        .authenticated())
+                // Cap body size pre-auth, then bind the validated token to a Stirling user after
+                // the bearer filter.
+                .addFilterBefore(
+                        new McpRequestSizeFilter(
+                                applicationProperties.getMcp().getMaxRequestBytes()),
+                        BearerTokenAuthenticationFilter.class)
+                .addFilterAfter(
+                        new McpUserBindingFilter(
+                                userService,
+                                auth.getUsernameClaim(),
+                                auth.isRequireExistingAccount()),
+                        BearerTokenAuthenticationFilter.class)
+                .oauth2ResourceServer(
+                        oauth2 ->
+                                oauth2.authenticationEntryPoint(
+                                                new McpAuthenticationEntryPoint(metadataPath))
+                                        // RFC 9728 protected-resource metadata for OAuth discovery.
+                                        .protectedResourceMetadata(
+                                                prm ->
+                                                        prm.protectedResourceMetadataCustomizer(
+                                                                builder -> {
+                                                                    if (!auth.getResourceId()
+                                                                            .isBlank()) {
+                                                                        builder.resource(
+                                                                                auth
+                                                                                        .getResourceId());
+                                                                    }
+                                                                    if (!auth.getIssuerUri()
+                                                                            .isBlank()) {
+                                                                        builder.authorizationServer(
+                                                                                auth
+                                                                                        .getIssuerUri());
+                                                                    }
+                                                                    builder.scope("mcp.tools.read");
+                                                                    builder.scope(
+                                                                            "mcp.tools.write");
+                                                                }))
+                                        .jwt(
+                                                jwt ->
+                                                        jwt.decoder(mcpJwtDecoder)
+                                                                .jwtAuthenticationConverter(
+                                                                        mcpJwtAuthenticationConverter())));
+        return http.build();
+    }
+
+    @Bean
+    JwtDecoder mcpJwtDecoder() {
+        ApplicationProperties.Mcp.Auth auth = applicationProperties.getMcp().getAuth();
+        if (auth.getIssuerUri().isBlank()) {
+            // Fail-closed decoder: rejects every token until the issuer is set.
+            return token -> {
+                throw new org.springframework.security.oauth2.jwt.BadJwtException(
+                        "mcp.auth.issuer-uri is not configured");
+            };
+        }
+        String jwksUri = auth.getJwksUri();
+        NimbusJwtDecoder decoder =
+                jwksUri.isBlank()
+                        ? NimbusJwtDecoder.withIssuerLocation(auth.getIssuerUri()).build()
+                        : NimbusJwtDecoder.withJwkSetUri(jwksUri).build();
+        OAuth2TokenValidator<Jwt> defaultValidators =
+                JwtValidators.createDefaultWithIssuer(auth.getIssuerUri());
+        OAuth2TokenValidator<Jwt> combined =
+                new DelegatingOAuth2TokenValidator<>(
+                        defaultValidators, new McpAudienceValidator(auth.getResourceId()));
+        decoder.setJwtValidator(combined);
+        return decoder;
+    }
+
+    private Converter<Jwt, AbstractAuthenticationToken> mcpJwtAuthenticationConverter() {
+        JwtGrantedAuthoritiesConverter scopes = new JwtGrantedAuthoritiesConverter();
+        scopes.setAuthorityPrefix("SCOPE_");
+        scopes.setAuthoritiesClaimName("scope");
+        JwtAuthenticationConverter converter = new JwtAuthenticationConverter();
+        converter.setJwtGrantedAuthoritiesConverter(
+                jwt -> {
+                    Collection<GrantedAuthority> out = new ArrayList<>(scopes.convert(jwt));
+                    List<String> aud = jwt.getAudience();
+                    if (aud != null) {
+                        for (String a : aud) {
+                            out.add(new SimpleGrantedAuthority("AUDIENCE_" + a));
+                        }
+                    }
+                    return out;
+                });
+        return converter;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/security/McpUserBindingFilter.java

@@ -0,0 +1,115 @@
+package stirling.software.proprietary.mcp.security;
+
+import java.io.IOException;
+import java.util.Optional;
+
+import org.springframework.security.authentication.UsernamePasswordAuthenticationToken;
+import org.springframework.security.core.Authentication;
+import org.springframework.security.core.context.SecurityContext;
+import org.springframework.security.core.context.SecurityContextHolder;
+import org.springframework.security.oauth2.jwt.Jwt;
+import org.springframework.security.oauth2.server.resource.authentication.JwtAuthenticationToken;
+import org.springframework.web.filter.OncePerRequestFilter;
+
+import jakarta.servlet.FilterChain;
+import jakarta.servlet.ServletException;
+import jakarta.servlet.http.HttpServletRequest;
+import jakarta.servlet.http.HttpServletResponse;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.proprietary.security.model.User;
+import stirling.software.proprietary.security.service.UserService;
+
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Binds an MCP-validated JWT to a provisioned Stirling user: optionally rejects subjects with no
+ * enabled account, then rebinds the principal to the canonical Stirling username (scope authorities
+ * only) so audit/metering attribute correctly.
+ */
+@Slf4j
+public class McpUserBindingFilter extends OncePerRequestFilter {
+
+    private static final ObjectMapper MAPPER = new ObjectMapper();
+
+    private final UserService userService;
+    private final String usernameClaim;
+    private final boolean requireExistingAccount;
+
+    public McpUserBindingFilter(
+            UserService userService, String usernameClaim, boolean requireExistingAccount) {
+        this.userService = userService;
+        this.usernameClaim =
+                (usernameClaim == null || usernameClaim.isBlank()) ? "sub" : usernameClaim;
+        this.requireExistingAccount = requireExistingAccount;
+    }
+
+    @Override
+    protected void doFilterInternal(
+            HttpServletRequest request, HttpServletResponse response, FilterChain filterChain)
+            throws ServletException, IOException {
+        Authentication current = SecurityContextHolder.getContext().getAuthentication();
+
+        // Only act on a JWT-authenticated request; everything else passes through.
+        if (current instanceof JwtAuthenticationToken jwtAuth && jwtAuth.isAuthenticated()) {
+            Jwt jwt = jwtAuth.getToken();
+            String username = jwt.getClaimAsString(usernameClaim);
+
+            if (username == null || username.isBlank()) {
+                reject(
+                        response,
+                        "Token is missing the '"
+                                + usernameClaim
+                                + "' claim used to map to a"
+                                + " Stirling user.");
+                return;
+            }
+
+            // Prefer the canonical username from the account record; fall back to the claim when
+            // binding is off.
+            String boundUsername = username;
+            if (requireExistingAccount) {
+                Optional<User> account = userService.findByUsernameIgnoreCase(username);
+                if (account.isEmpty() || !account.get().isEnabled()) {
+                    log.warn(
+                            "MCP access denied: token subject '{}' has no active Stirling account",
+                            sanitizeForLog(username));
+                    reject(
+                            response,
+                            "MCP access requires a provisioned, enabled Stirling account for this"
+                                    + " subject.");
+                    return;
+                }
+                boundUsername = account.get().getUsername();
+            }
+
+            // Rebind to the Stirling username, carrying only the OAuth scope authorities.
+            UsernamePasswordAuthenticationToken bound =
+                    new UsernamePasswordAuthenticationToken(
+                            boundUsername, null, jwtAuth.getAuthorities());
+            bound.setDetails(jwtAuth.getDetails());
+            SecurityContext context = SecurityContextHolder.createEmptyContext();
+            context.setAuthentication(bound);
+            SecurityContextHolder.setContext(context);
+        }
+
+        filterChain.doFilter(request, response);
+    }
+
+    /** Strip CR/LF so a crafted claim value can't forge log lines. */
+    private static String sanitizeForLog(String value) {
+        return value == null ? null : value.replace('\r', ' ').replace('\n', ' ');
+    }
+
+    private void reject(HttpServletResponse response, String message) throws IOException {
+        SecurityContextHolder.clearContext();
+        response.setStatus(HttpServletResponse.SC_FORBIDDEN);
+        response.setContentType("application/json");
+        ObjectNode body = MAPPER.createObjectNode();
+        body.put("error", "insufficient_account");
+        body.put("message", message);
+        response.getWriter().write(MAPPER.writeValueAsString(body));
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/AbstractCategoryTool.java

@@ -0,0 +1,154 @@
+package stirling.software.proprietary.mcp.tools;
+
+import java.util.List;
+
+import org.springframework.beans.factory.ObjectProvider;
+
+import stirling.software.proprietary.mcp.McpCallContext;
+import stirling.software.proprietary.mcp.McpTool;
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationCategory;
+import stirling.software.proprietary.mcp.catalog.OperationMeta;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ArrayNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Common scaffolding for the PDF category tools. Operation ids and summaries come from the live
+ * {@link McpToolCatalog}.
+ */
+abstract class AbstractCategoryTool implements McpTool {
+
+    protected final ObjectMapper mapper;
+    protected final ObjectProvider<McpToolCatalog> catalogProvider;
+    protected final ObjectProvider<McpOperationExecutor> executorProvider;
+
+    protected AbstractCategoryTool(
+            ObjectMapper mapper,
+            ObjectProvider<McpToolCatalog> catalog,
+            ObjectProvider<McpOperationExecutor> executor) {
+        this.mapper = mapper;
+        this.catalogProvider = catalog;
+        this.executorProvider = executor;
+    }
+
+    protected abstract OperationCategory category();
+
+    protected List<OperationMeta> enabledOperations() {
+        McpToolCatalog catalog = catalogProvider.getIfAvailable();
+        if (catalog == null) {
+            return List.of();
+        }
+        return catalog.enabledOps(category());
+    }
+
+    @Override
+    public ObjectNode inputSchema() {
+        ObjectNode schema = mapper.createObjectNode();
+        schema.put("type", "object");
+        schema.put("additionalProperties", false);
+
+        ObjectNode props = schema.putObject("properties");
+
+        ObjectNode op = props.putObject("operation");
+        op.put("type", "string");
+        List<OperationMeta> enabled = enabledOperations();
+        StringBuilder opDesc = new StringBuilder();
+        opDesc.append(
+                "Operation id from this category. Call stirling_describe_operation first to learn"
+                        + " the exact parameters schema. Available operations:\n");
+        ArrayNode opEnum = op.putArray("enum");
+        for (OperationMeta m : enabled) {
+            opEnum.add(m.id());
+            opDesc.append("- ").append(m.id()).append(" - ").append(m.summary()).append('\n');
+        }
+        op.put("description", opDesc.toString().trim());
+
+        ObjectNode params = props.putObject("parameters");
+        params.put("type", "object");
+        params.put(
+                "description",
+                "Per-operation parameters. Schema available via stirling_describe_operation.");
+        params.put("additionalProperties", true);
+
+        McpToolSupport.stringProperty(
+                props,
+                "file",
+                "Base64-encoded file content to process. The recommended way to provide a file for"
+                        + " most uses. Bounded by the MCP request size limit; for very large files"
+                        + " use 'fileId' instead.");
+        McpToolSupport.stringProperty(
+                props,
+                "fileName",
+                "Optional original filename (with extension) for the input; helps operations that"
+                        + " key off file type.");
+        McpToolSupport.stringProperty(
+                props,
+                "fileId",
+                "Reference to a file already stored via stirling_upload. Recommended only for large"
+                        + " files or multi-step workflows; most users should pass the file inline"
+                        + " via 'file' instead.");
+
+        ArrayNode required = schema.putArray("required");
+        required.add("operation");
+        return schema;
+    }
+
+    @Override
+    public ObjectNode call(JsonNode arguments, McpCallContext context) {
+        JsonNode opNode = arguments == null ? null : arguments.get("operation");
+        // No operation chosen: return this category's operation list.
+        if (opNode == null || !opNode.isTextual() || opNode.asText().isBlank()) {
+            return operationListError(null);
+        }
+        String opId = opNode.asText();
+        McpToolCatalog catalog = catalogProvider.getIfAvailable();
+        if (catalog == null) {
+            return McpResponses.error(mapper, "MCP catalog is not available");
+        }
+        OperationMeta meta = catalog.findByOperationId(opId).orElse(null);
+        // Invalid/disabled/wrong-category op: return this category's operations.
+        if (meta == null || meta.category() != category()) {
+            return operationListError(opId);
+        }
+        if (!context.hasScope(meta.requiredScope())) {
+            return McpResponses.error(
+                    mapper,
+                    "Insufficient scope: this operation requires '" + meta.requiredScope() + "'.");
+        }
+        McpOperationExecutor executor = executorProvider.getIfAvailable();
+        if (executor == null) {
+            return McpResponses.error(mapper, "MCP execution is not available.");
+        }
+        return executor.execute(meta, arguments);
+    }
+
+    /**
+     * Error for a missing/unknown operation, listing this category's available operation ids and
+     * summaries.
+     */
+    private ObjectNode operationListError(String badOpId) {
+        StringBuilder sb = new StringBuilder();
+        if (badOpId == null) {
+            sb.append("Missing required argument 'operation' for ").append(category().toolName());
+        } else {
+            sb.append("Unknown or disabled operation '")
+                    .append(badOpId)
+                    .append("' for ")
+                    .append(category().toolName());
+        }
+        List<OperationMeta> ops = enabledOperations();
+        if (ops.isEmpty()) {
+            sb.append(". No operations are currently available in this category.");
+        } else {
+            sb.append(". Available operations:");
+            for (OperationMeta m : ops) {
+                sb.append("\n- ").append(m.id()).append(" - ").append(m.summary());
+            }
+            sb.append("\nRe-call this tool with a valid 'operation'.");
+        }
+        return McpResponses.error(mapper, sb.toString());
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/DescribeOperationTool.java

@@ -0,0 +1,84 @@
+package stirling.software.proprietary.mcp.tools;
+
+import org.springframework.beans.factory.ObjectProvider;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.stereotype.Component;
+
+import stirling.software.proprietary.mcp.McpCallContext;
+import stirling.software.proprietary.mcp.McpTool;
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationMeta;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ArrayNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/** Returns the JSON Schema for one operation's parameters, from the live {@link McpToolCatalog}. */
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class DescribeOperationTool implements McpTool {
+
+    private final ObjectMapper mapper;
+    private final ObjectProvider<McpToolCatalog> catalogProvider;
+
+    public DescribeOperationTool(ObjectMapper mapper, ObjectProvider<McpToolCatalog> catalog) {
+        this.mapper = mapper;
+        this.catalogProvider = catalog;
+    }
+
+    @Override
+    public String name() {
+        return "stirling_describe_operation";
+    }
+
+    @Override
+    public String description() {
+        return "Return the full JSON Schema for one Stirling operation's parameters. Call this "
+                + "before invoking a category tool to learn the exact shape of `parameters`. "
+                + "Argument: { operation: <op-id> } where <op-id> appears in the enum of any "
+                + "category tool (stirling_convert, _pages, _misc, _security, _ai).";
+    }
+
+    @Override
+    public ObjectNode inputSchema() {
+        ObjectNode schema = mapper.createObjectNode();
+        schema.put("type", "object");
+        schema.put("additionalProperties", false);
+        ObjectNode props = schema.putObject("properties");
+        ObjectNode op = props.putObject("operation");
+        op.put("type", "string");
+        op.put(
+                "description",
+                "Operation id (e.g. compress-pdf, pdf-to-word, q-and-a). See category tool enums.");
+        ArrayNode required = schema.putArray("required");
+        required.add("operation");
+        return schema;
+    }
+
+    @Override
+    public ObjectNode call(JsonNode arguments, McpCallContext context) {
+        JsonNode opNode = arguments == null ? null : arguments.get("operation");
+        if (opNode == null || !opNode.isTextual() || opNode.asText().isBlank()) {
+            return McpResponses.error(mapper, "Missing required argument: operation");
+        }
+        String opId = opNode.asText();
+        McpToolCatalog catalog = catalogProvider.getIfAvailable();
+        if (catalog == null) {
+            return McpResponses.error(mapper, "MCP catalog is not available");
+        }
+        OperationMeta meta = catalog.findByOperationId(opId).orElse(null);
+        if (meta == null) {
+            return McpResponses.error(mapper, "Unknown or disabled operation: " + opId);
+        }
+
+        ObjectNode payload = mapper.createObjectNode();
+        payload.put("operation", meta.id());
+        payload.put("category", meta.category().toolName());
+        payload.put("summary", meta.summary());
+        payload.put("endpoint", meta.endpointPath());
+        payload.put("requiredScope", meta.requiredScope());
+        payload.set("parametersSchema", meta.paramSchema());
+        return McpResponses.json(mapper, payload);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/McpOperationExecutor.java

@@ -0,0 +1,255 @@
+package stirling.software.proprietary.mcp.tools;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.charset.StandardCharsets;
+import java.util.Base64;
+import java.util.List;
+import java.util.Map;
+
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.core.io.ByteArrayResource;
+import org.springframework.core.io.Resource;
+import org.springframework.http.MediaType;
+import org.springframework.http.ResponseEntity;
+import org.springframework.stereotype.Component;
+import org.springframework.util.LinkedMultiValueMap;
+import org.springframework.util.MultiValueMap;
+import org.springframework.web.client.RestClientResponseException;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.service.FileStorage;
+import stirling.software.common.service.InternalApiClient;
+import stirling.software.common.service.InternalApiTimeoutException;
+import stirling.software.proprietary.mcp.catalog.OperationMeta;
+
+import tools.jackson.core.type.TypeReference;
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Runs a JAVA_ENDPOINT operation: resolves the input file (inline base64 or a fileId), dispatches
+ * to the Stirling endpoint over the loopback via {@link InternalApiClient}, and stores the result.
+ */
+@Slf4j
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class McpOperationExecutor {
+
+    private final ObjectMapper mapper;
+    private final InternalApiClient internalApiClient;
+    private final FileStorage fileStorage;
+    private final ApplicationProperties applicationProperties;
+
+    public McpOperationExecutor(
+            ObjectMapper mapper,
+            InternalApiClient internalApiClient,
+            FileStorage fileStorage,
+            ApplicationProperties applicationProperties) {
+        this.mapper = mapper;
+        this.internalApiClient = internalApiClient;
+        this.fileStorage = fileStorage;
+        this.applicationProperties = applicationProperties;
+    }
+
+    public ObjectNode execute(OperationMeta meta, JsonNode arguments) {
+        String fileName = McpToolSupport.textArg(arguments, "fileName");
+        String fileId = McpToolSupport.textArg(arguments, "fileId");
+        byte[] inputBytes;
+        String inputName;
+        if (fileId != null) {
+            try {
+                if (!fileStorage.fileExists(fileId)) {
+                    return McpResponses.error(
+                            mapper,
+                            "Unknown or inaccessible fileId '"
+                                    + fileId
+                                    + "'. Re-upload with stirling_upload.");
+                }
+                inputBytes = fileStorage.retrieveBytes(fileId);
+            } catch (SecurityException e) {
+                return McpResponses.error(
+                        mapper,
+                        "Unknown or inaccessible fileId '"
+                                + fileId
+                                + "'. Re-upload with stirling_upload.");
+            } catch (IOException e) {
+                return McpResponses.error(mapper, "Could not read fileId '" + fileId + "'.");
+            }
+            inputName = fileName != null ? fileName : fileId;
+        } else {
+            String base64 = McpToolSupport.textArg(arguments, "file");
+            if (base64 == null) {
+                return McpResponses.error(
+                        mapper,
+                        "This operation needs an input file. Pass 'file' as base64 (recommended for"
+                                + " most files), or 'fileId' from stirling_upload for large files.");
+            }
+            inputBytes = McpToolSupport.decodeBase64OrNull(base64);
+            if (inputBytes == null) {
+                return McpResponses.error(mapper, "The 'file' argument is not valid base64.");
+            }
+            inputName = fileName != null ? fileName : "input.pdf";
+        }
+
+        MultiValueMap<String, Object> body = new LinkedMultiValueMap<>();
+        body.add("fileInput", bytesResource(inputBytes, inputName));
+        addParameters(body, arguments == null ? null : arguments.get("parameters"));
+
+        ResponseEntity<Resource> response;
+        try {
+            response = internalApiClient.post(meta.endpointPath(), body);
+        } catch (InternalApiTimeoutException e) {
+            return McpResponses.error(
+                    mapper,
+                    meta.id()
+                            + " timed out after "
+                            + e.getReadTimeout().toSeconds()
+                            + "s. Try a smaller file or a different approach.");
+        } catch (RestClientResponseException e) {
+            log.warn(
+                    "MCP {} upstream error: HTTP {} - {}",
+                    meta.id(),
+                    e.getStatusCode().value(),
+                    snippet(e.getResponseBodyAsString()));
+            return McpResponses.error(
+                    mapper, meta.id() + " failed: HTTP " + e.getStatusCode().value() + ".");
+        } catch (SecurityException e) {
+            return McpResponses.error(
+                    mapper, meta.id() + " endpoint is not permitted for MCP dispatch.");
+        } catch (RuntimeException e) {
+            log.warn("MCP execution of {} failed", meta.id(), e);
+            return McpResponses.error(
+                    mapper, meta.id() + " failed unexpectedly. See server logs for details.");
+        }
+        return buildResult(meta, response);
+    }
+
+    private ObjectNode buildResult(OperationMeta meta, ResponseEntity<Resource> response) {
+        Resource body = response.getBody();
+        if (body == null) {
+            return McpResponses.error(mapper, meta.id() + " returned an empty response.");
+        }
+        MediaType contentType = response.getHeaders().getContentType();
+
+        // A JSON body is a structured report (e.g. get-info), not a file.
+        if (contentType != null && MediaType.APPLICATION_JSON.isCompatibleWith(contentType)) {
+            try (InputStream is = body.getInputStream()) {
+                return McpResponses.text(
+                        mapper, new String(is.readAllBytes(), StandardCharsets.UTF_8));
+            } catch (IOException e) {
+                return McpResponses.error(mapper, "Failed to read " + meta.id() + " result.");
+            }
+        }
+
+        String filename =
+                body.getFilename() == null || body.getFilename().isBlank()
+                        ? meta.id()
+                        : body.getFilename();
+        String mimeType =
+                contentType != null
+                        ? contentType.toString()
+                        : MediaType.APPLICATION_OCTET_STREAM_VALUE;
+        long maxInline = applicationProperties.getMcp().getMaxInlineResponseBytes();
+        try {
+            long size = body.contentLength();
+            byte[] inline = null;
+            if (size >= 0 && size <= maxInline) {
+                try (InputStream is = body.getInputStream()) {
+                    inline = is.readAllBytes();
+                }
+            }
+            String fileId =
+                    inline != null
+                            ? fileStorage.storeBytes(inline, filename)
+                            : storeStreamed(body, filename);
+            String summary =
+                    meta.id()
+                            + " succeeded. Result: "
+                            + filename
+                            + " ("
+                            + size
+                            + " bytes), fileId="
+                            + fileId
+                            + ". ";
+            if (inline != null) {
+                return McpResponses.result(
+                        mapper,
+                        false,
+                        McpResponses.textBlock(
+                                mapper, summary + "The file is included inline below."),
+                        McpResponses.resourceBlock(
+                                mapper,
+                                "stirling://file/" + fileId,
+                                mimeType,
+                                Base64.getEncoder().encodeToString(inline)));
+            }
+            return McpResponses.result(
+                    mapper,
+                    false,
+                    McpResponses.textBlock(
+                            mapper,
+                            summary
+                                    + "Large result - fetch it with stirling_download {\"fileId\":\""
+                                    + fileId
+                                    + "\"}, or pass this fileId to another operation."));
+        } catch (IOException e) {
+            return McpResponses.error(mapper, "Failed to store " + meta.id() + " result.");
+        }
+    }
+
+    private String storeStreamed(Resource body, String filename) throws IOException {
+        try (InputStream is = body.getInputStream()) {
+            return fileStorage.storeInputStream(is, filename).fileId();
+        }
+    }
+
+    private void addParameters(MultiValueMap<String, Object> body, JsonNode params) {
+        if (params == null || !params.isObject()) {
+            return;
+        }
+        Map<String, Object> map =
+                mapper.convertValue(params, new TypeReference<Map<String, Object>>() {});
+        for (Map.Entry<String, Object> entry : map.entrySet()) {
+            Object value = entry.getValue();
+            if (value == null) {
+                continue;
+            }
+            if (value instanceof List<?> list) {
+                if (containsStructured(list)) {
+                    body.add(entry.getKey(), mapper.writeValueAsString(list));
+                } else {
+                    list.forEach(item -> body.add(entry.getKey(), item));
+                }
+            } else if (value instanceof Map<?, ?>) {
+                body.add(entry.getKey(), mapper.writeValueAsString(value));
+            } else {
+                body.add(entry.getKey(), value);
+            }
+        }
+    }
+
+    private static boolean containsStructured(List<?> list) {
+        return list.stream().anyMatch(item -> item instanceof Map<?, ?> || item instanceof List<?>);
+    }
+
+    private static Resource bytesResource(byte[] bytes, String filename) {
+        return new ByteArrayResource(bytes) {
+            @Override
+            public String getFilename() {
+                return filename;
+            }
+        };
+    }
+
+    private static String snippet(String body) {
+        if (body == null || body.isBlank()) {
+            return "(no body)";
+        }
+        String trimmed = body.strip();
+        return trimmed.length() > 300 ? trimmed.substring(0, 300) + "..." : trimmed;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/McpResponses.java

@@ -0,0 +1,80 @@
+package stirling.software.proprietary.mcp.tools;
+
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ArrayNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/** Helpers for the MCP {@code CallToolResult} response shape. */
+public final class McpResponses {
+
+    private McpResponses() {}
+
+    /** Plain-text content block. */
+    public static ObjectNode text(ObjectMapper mapper, String text) {
+        ObjectNode block = mapper.createObjectNode();
+        block.put("type", "text");
+        block.put("text", text);
+        return wrap(mapper, block, false);
+    }
+
+    /** Plain-text error ({@code isError:true}). */
+    public static ObjectNode error(ObjectMapper mapper, String message) {
+        ObjectNode block = mapper.createObjectNode();
+        block.put("type", "text");
+        block.put("text", message);
+        return wrap(mapper, block, true);
+    }
+
+    /** JSON payload as embedded text. */
+    public static ObjectNode json(ObjectMapper mapper, ObjectNode payload) {
+        ObjectNode block = mapper.createObjectNode();
+        block.put("type", "text");
+        block.put("text", payload.toString());
+        return wrap(mapper, block, false);
+    }
+
+    /** A text content block (unwrapped). */
+    public static ObjectNode textBlock(ObjectMapper mapper, String text) {
+        ObjectNode block = mapper.createObjectNode();
+        block.put("type", "text");
+        block.put("text", text);
+        return block;
+    }
+
+    /** An embedded-resource content block carrying base64 file content. */
+    public static ObjectNode resourceBlock(
+            ObjectMapper mapper, String uri, String mimeType, String base64) {
+        ObjectNode block = mapper.createObjectNode();
+        block.put("type", "resource");
+        ObjectNode res = block.putObject("resource");
+        res.put("uri", uri);
+        if (mimeType != null) {
+            res.put("mimeType", mimeType);
+        }
+        res.put("blob", base64);
+        return block;
+    }
+
+    /** Build a result from explicit content blocks. */
+    public static ObjectNode result(ObjectMapper mapper, boolean isError, ObjectNode... blocks) {
+        ObjectNode result = mapper.createObjectNode();
+        ArrayNode content = result.putArray("content");
+        for (ObjectNode b : blocks) {
+            content.add(b);
+        }
+        if (isError) {
+            result.put("isError", true);
+        }
+        return result;
+    }
+
+    private static ObjectNode wrap(ObjectMapper mapper, ObjectNode block, boolean isError) {
+        ObjectNode result = mapper.createObjectNode();
+        ArrayNode content = result.putArray("content");
+        content.add(block);
+        if (isError) {
+            result.put("isError", true);
+        }
+        return result;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/McpToolSupport.java

@@ -0,0 +1,43 @@
+package stirling.software.proprietary.mcp.tools;
+
+import java.util.Base64;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/** Shared helpers for MCP tools: argument parsing and JSON-Schema building. */
+final class McpToolSupport {
+
+    private McpToolSupport() {}
+
+    /** Trimmed text value of an argument, or null if absent, blank, or not a string. */
+    static String textArg(JsonNode args, String field) {
+        if (args == null) {
+            return null;
+        }
+        JsonNode node = args.get(field);
+        if (node == null || !node.isTextual()) {
+            return null;
+        }
+        String value = node.asText().trim();
+        return value.isEmpty() ? null : value;
+    }
+
+    /** Decode base64 content, or null if the input is not valid base64. */
+    static byte[] decodeBase64OrNull(String base64) {
+        try {
+            return Base64.getDecoder().decode(base64);
+        } catch (IllegalArgumentException e) {
+            return null;
+        }
+    }
+
+    /**
+     * Add a {@code string} property with a description to a JSON-Schema {@code properties} node.
+     */
+    static void stringProperty(ObjectNode properties, String name, String description) {
+        ObjectNode prop = properties.putObject(name);
+        prop.put("type", "string");
+        prop.put("description", description);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/StirlingAiTool.java

@@ -0,0 +1,149 @@
+package stirling.software.proprietary.mcp.tools;
+
+import java.io.IOException;
+import java.util.List;
+
+import org.springframework.beans.factory.ObjectProvider;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.stereotype.Component;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.proprietary.mcp.McpCallContext;
+import stirling.software.proprietary.mcp.McpTool;
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationCategory;
+import stirling.software.proprietary.mcp.catalog.OperationMeta;
+import stirling.software.proprietary.service.AiEngineClient;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ArrayNode;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Exposes curated Python agent capabilities as a single MCP tool, sourced from the engine
+ * capabilities manifest.
+ */
+@Slf4j
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class StirlingAiTool implements McpTool {
+
+    private final ObjectMapper mapper;
+    private final ObjectProvider<McpToolCatalog> catalogProvider;
+    private final ObjectProvider<AiEngineClient> engineClientProvider;
+
+    public StirlingAiTool(
+            ObjectMapper mapper,
+            ObjectProvider<McpToolCatalog> catalog,
+            ObjectProvider<AiEngineClient> engineClient) {
+        this.mapper = mapper;
+        this.catalogProvider = catalog;
+        this.engineClientProvider = engineClient;
+    }
+
+    @Override
+    public String name() {
+        return "stirling_ai";
+    }
+
+    @Override
+    public String description() {
+        return "Invoke a Stirling AI agent capability (Q&A about a PDF, edit-plan generation,"
+                + " inline comments, math audit, draft-spec helper). Call"
+                + " stirling_describe_operation with the chosen capability id to get its"
+                + " parameters schema before invoking this tool. Some capabilities return content"
+                + " inline; others return a job reference that resolves to a file when ready.";
+    }
+
+    @Override
+    public ObjectNode inputSchema() {
+        ObjectNode schema = mapper.createObjectNode();
+        schema.put("type", "object");
+        schema.put("additionalProperties", false);
+        ObjectNode props = schema.putObject("properties");
+
+        ObjectNode op = props.putObject("operation");
+        op.put("type", "string");
+        StringBuilder desc = new StringBuilder();
+        desc.append("Capability id from the engine manifest. Available capabilities:\n");
+        ArrayNode opEnum = op.putArray("enum");
+        for (OperationMeta m : aiOps()) {
+            opEnum.add(m.id());
+            desc.append("- ").append(m.id()).append(" - ").append(m.summary()).append('\n');
+        }
+        op.put("description", desc.toString().trim());
+
+        ObjectNode params = props.putObject("parameters");
+        params.put("type", "object");
+        params.put("description", "Per-capability parameters.");
+        params.put("additionalProperties", true);
+
+        ObjectNode fileId = props.putObject("fileId");
+        fileId.put("type", "string");
+        fileId.put(
+                "description",
+                "Reference to a previously-uploaded PDF in Stirling's job store. Required for"
+                        + " capabilities that consume a document.");
+
+        ArrayNode required = schema.putArray("required");
+        required.add("operation");
+        return schema;
+    }
+
+    @Override
+    public ObjectNode call(JsonNode arguments, McpCallContext context) {
+        JsonNode opNode = arguments == null ? null : arguments.get("operation");
+        if (opNode == null || !opNode.isTextual() || opNode.asText().isBlank()) {
+            return McpResponses.error(mapper, "Missing required argument: operation");
+        }
+        String opId = opNode.asText();
+        McpToolCatalog catalog = catalogProvider.getIfAvailable();
+        if (catalog == null) {
+            return McpResponses.error(mapper, "MCP catalog is not available");
+        }
+        OperationMeta meta = catalog.findByOperationId(opId).orElse(null);
+        if (meta == null || meta.category() != OperationCategory.AI) {
+            return McpResponses.error(
+                    mapper,
+                    "Unknown AI capability '"
+                            + opId
+                            + "'. The engine manifest may not be loaded yet - retry shortly or"
+                            + " confirm the engine is reachable.");
+        }
+        if (!context.hasScope(meta.requiredScope())) {
+            return McpResponses.error(
+                    mapper,
+                    "Insufficient scope: this capability requires '" + meta.requiredScope() + "'.");
+        }
+        AiEngineClient client = engineClientProvider.getIfAvailable();
+        if (client == null) {
+            return McpResponses.error(
+                    mapper, "AI engine client is not configured - enable aiEngine in settings.");
+        }
+        if (meta.endpointPath() == null) {
+            return McpResponses.error(
+                    mapper,
+                    "Capability '" + opId + "' has no route configured in the engine manifest.");
+        }
+        JsonNode params = arguments.get("parameters");
+        String body = (params == null ? mapper.createObjectNode() : params).toString();
+        try {
+            String response = client.post(meta.endpointPath(), body, context.stirlingUserId());
+            return McpResponses.text(mapper, response);
+        } catch (IOException e) {
+            log.warn("MCP AI capability '{}' engine request failed", opId, e);
+            return McpResponses.error(
+                    mapper, "Engine request failed for capability '" + opId + "'.");
+        }
+    }
+
+    private List<OperationMeta> aiOps() {
+        McpToolCatalog catalog = catalogProvider.getIfAvailable();
+        if (catalog == null) {
+            return List.of();
+        }
+        return catalog.enabledOps(OperationCategory.AI);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/StirlingConvertTool.java

@@ -0,0 +1,41 @@
+package stirling.software.proprietary.mcp.tools;
+
+import org.springframework.beans.factory.ObjectProvider;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.stereotype.Component;
+
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationCategory;
+
+import tools.jackson.databind.ObjectMapper;
+
+/** Exposes the {@code /api/v1/convert/*} namespace as a single MCP tool. */
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class StirlingConvertTool extends AbstractCategoryTool {
+
+    public StirlingConvertTool(
+            ObjectMapper mapper,
+            ObjectProvider<McpToolCatalog> catalog,
+            ObjectProvider<McpOperationExecutor> executor) {
+        super(mapper, catalog, executor);
+    }
+
+    @Override
+    public String name() {
+        return "stirling_convert";
+    }
+
+    @Override
+    public String description() {
+        return "Convert files between PDF and other formats (PDF<->Word, PDF<->image, HTML->PDF,"
+                + " etc.). Inspect the `operation` enum, then call stirling_describe_operation"
+                + " with the chosen op to get its parameters JSON Schema before calling this"
+                + " tool.";
+    }
+
+    @Override
+    protected OperationCategory category() {
+        return OperationCategory.CONVERT;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/StirlingDownloadTool.java

@@ -0,0 +1,113 @@
+package stirling.software.proprietary.mcp.tools;
+
+import java.io.IOException;
+import java.util.Base64;
+
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.http.MediaType;
+import org.springframework.stereotype.Component;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.service.FileStorage;
+import stirling.software.proprietary.mcp.McpCallContext;
+import stirling.software.proprietary.mcp.McpTool;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Fetches a stored file's content by fileId, returned inline as base64. For large results that were
+ * not returned inline by an operation.
+ */
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class StirlingDownloadTool implements McpTool {
+
+    private final ObjectMapper mapper;
+    private final FileStorage fileStorage;
+    private final ApplicationProperties applicationProperties;
+
+    public StirlingDownloadTool(
+            ObjectMapper mapper,
+            FileStorage fileStorage,
+            ApplicationProperties applicationProperties) {
+        this.mapper = mapper;
+        this.fileStorage = fileStorage;
+        this.applicationProperties = applicationProperties;
+    }
+
+    @Override
+    public String name() {
+        return "stirling_download";
+    }
+
+    @Override
+    public String description() {
+        return "Fetch a stored file's content by fileId (e.g. an operation result), returned inline"
+                + " as base64. Recommended only when a result was too large to be returned inline."
+                + " Argument: { fileId: <id> }.";
+    }
+
+    @Override
+    public ObjectNode inputSchema() {
+        ObjectNode schema = mapper.createObjectNode();
+        schema.put("type", "object");
+        schema.put("additionalProperties", false);
+        ObjectNode props = schema.putObject("properties");
+        McpToolSupport.stringProperty(
+                props, "fileId", "Id of a stored file (e.g. an operation result's fileId).");
+        schema.putArray("required").add("fileId");
+        return schema;
+    }
+
+    @Override
+    public ObjectNode call(JsonNode arguments, McpCallContext context) {
+        if (!context.hasScope("mcp.tools.read")) {
+            return McpResponses.error(
+                    mapper, "Insufficient scope: stirling_download requires 'mcp.tools.read'.");
+        }
+        String fileId = McpToolSupport.textArg(arguments, "fileId");
+        if (fileId == null) {
+            return McpResponses.error(mapper, "Missing required argument: fileId.");
+        }
+        long maxInline = applicationProperties.getMcp().getMaxInlineResponseBytes();
+        try {
+            if (!fileStorage.fileExists(fileId)) {
+                return McpResponses.error(
+                        mapper, "Unknown or inaccessible fileId '" + fileId + "'.");
+            }
+            long size = fileStorage.getFileSize(fileId);
+            if (size > maxInline) {
+                return McpResponses.error(
+                        mapper,
+                        "File is "
+                                + size
+                                + " bytes, over the inline limit of "
+                                + maxInline
+                                + " bytes. Raise mcp.maxInlineResponseBytes or retrieve it via the"
+                                + " Stirling UI/API.");
+            }
+            byte[] bytes = fileStorage.retrieveBytes(fileId);
+            return McpResponses.result(
+                    mapper,
+                    false,
+                    McpResponses.textBlock(
+                            mapper,
+                            "File "
+                                    + fileId
+                                    + " ("
+                                    + bytes.length
+                                    + " bytes) included inline below."),
+                    McpResponses.resourceBlock(
+                            mapper,
+                            "stirling://file/" + fileId,
+                            MediaType.APPLICATION_OCTET_STREAM_VALUE,
+                            Base64.getEncoder().encodeToString(bytes)));
+        } catch (SecurityException e) {
+            return McpResponses.error(mapper, "Unknown or inaccessible fileId '" + fileId + "'.");
+        } catch (IOException e) {
+            return McpResponses.error(mapper, "Failed to read fileId '" + fileId + "'.");
+        }
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/StirlingMiscTool.java

@@ -0,0 +1,40 @@
+package stirling.software.proprietary.mcp.tools;
+
+import org.springframework.beans.factory.ObjectProvider;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.stereotype.Component;
+
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationCategory;
+
+import tools.jackson.databind.ObjectMapper;
+
+/** Exposes the {@code /api/v1/misc/*} namespace as a single MCP tool. */
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class StirlingMiscTool extends AbstractCategoryTool {
+
+    public StirlingMiscTool(
+            ObjectMapper mapper,
+            ObjectProvider<McpToolCatalog> catalog,
+            ObjectProvider<McpOperationExecutor> executor) {
+        super(mapper, catalog, executor);
+    }
+
+    @Override
+    public String name() {
+        return "stirling_misc";
+    }
+
+    @Override
+    public String description() {
+        return "Miscellaneous PDF operations: compress, OCR, stamp / watermark, edit metadata,"
+                + " flatten, repair, and similar utilities. Call stirling_describe_operation with"
+                + " the chosen op to get its parameters schema before invoking this tool.";
+    }
+
+    @Override
+    protected OperationCategory category() {
+        return OperationCategory.MISC;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/StirlingPagesTool.java

@@ -0,0 +1,40 @@
+package stirling.software.proprietary.mcp.tools;
+
+import org.springframework.beans.factory.ObjectProvider;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.stereotype.Component;
+
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationCategory;
+
+import tools.jackson.databind.ObjectMapper;
+
+/** Exposes the {@code /api/v1/general/*} (page operations) namespace as a single MCP tool. */
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class StirlingPagesTool extends AbstractCategoryTool {
+
+    public StirlingPagesTool(
+            ObjectMapper mapper,
+            ObjectProvider<McpToolCatalog> catalog,
+            ObjectProvider<McpOperationExecutor> executor) {
+        super(mapper, catalog, executor);
+    }
+
+    @Override
+    public String name() {
+        return "stirling_pages";
+    }
+
+    @Override
+    public String description() {
+        return "Manipulate PDF pages: merge, split, rotate, rearrange, crop, delete, overlay,"
+                + " add blank pages. Call stirling_describe_operation with the chosen op to get"
+                + " its parameters schema before invoking this tool.";
+    }
+
+    @Override
+    protected OperationCategory category() {
+        return OperationCategory.PAGES;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/StirlingSecurityTool.java

@@ -0,0 +1,41 @@
+package stirling.software.proprietary.mcp.tools;
+
+import org.springframework.beans.factory.ObjectProvider;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.stereotype.Component;
+
+import stirling.software.proprietary.mcp.catalog.McpToolCatalog;
+import stirling.software.proprietary.mcp.catalog.OperationCategory;
+
+import tools.jackson.databind.ObjectMapper;
+
+/** Exposes the {@code /api/v1/security/*} namespace as a single MCP tool. */
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class StirlingSecurityTool extends AbstractCategoryTool {
+
+    public StirlingSecurityTool(
+            ObjectMapper mapper,
+            ObjectProvider<McpToolCatalog> catalog,
+            ObjectProvider<McpOperationExecutor> executor) {
+        super(mapper, catalog, executor);
+    }
+
+    @Override
+    public String name() {
+        return "stirling_security";
+    }
+
+    @Override
+    public String description() {
+        return "Security-related PDF operations: password add/remove, redact, sanitize, certify"
+                + " / sign with cert, validate signature, add watermark. Call"
+                + " stirling_describe_operation with the chosen op to get its parameters schema"
+                + " before invoking this tool.";
+    }
+
+    @Override
+    protected OperationCategory category() {
+        return OperationCategory.SECURITY;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/mcp/tools/StirlingUploadTool.java

@@ -0,0 +1,96 @@
+package stirling.software.proprietary.mcp.tools;
+
+import java.io.IOException;
+
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.stereotype.Component;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.service.FileStorage;
+import stirling.software.proprietary.mcp.McpCallContext;
+import stirling.software.proprietary.mcp.McpTool;
+
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.node.ObjectNode;
+
+/**
+ * Stores a file server-side and returns a fileId. For large files or multi-step workflows only -
+ * most operations accept the file inline via their {@code file} argument.
+ */
+@Slf4j
+@Component
+@ConditionalOnProperty(name = "mcp.enabled", havingValue = "true")
+public class StirlingUploadTool implements McpTool {
+
+    private final ObjectMapper mapper;
+    private final FileStorage fileStorage;
+
+    public StirlingUploadTool(ObjectMapper mapper, FileStorage fileStorage) {
+        this.mapper = mapper;
+        this.fileStorage = fileStorage;
+    }
+
+    @Override
+    public String name() {
+        return "stirling_upload";
+    }
+
+    @Override
+    public String description() {
+        return "Store a file server-side and get back a fileId to reuse across operations."
+                + " Recommended only for large files or multi-step workflows; for a single"
+                + " operation on a typical file, pass the file inline via the operation's `file`"
+                + " argument instead. Argument: { file: <base64>, fileName?: <name> }.";
+    }
+
+    @Override
+    public ObjectNode inputSchema() {
+        ObjectNode schema = mapper.createObjectNode();
+        schema.put("type", "object");
+        schema.put("additionalProperties", false);
+        ObjectNode props = schema.putObject("properties");
+        McpToolSupport.stringProperty(props, "file", "Base64-encoded file content.");
+        McpToolSupport.stringProperty(
+                props, "fileName", "Optional original filename (with extension).");
+        schema.putArray("required").add("file");
+        return schema;
+    }
+
+    @Override
+    public ObjectNode call(JsonNode arguments, McpCallContext context) {
+        if (!context.hasScope("mcp.tools.write")) {
+            return McpResponses.error(
+                    mapper, "Insufficient scope: stirling_upload requires 'mcp.tools.write'.");
+        }
+        String base64 = McpToolSupport.textArg(arguments, "file");
+        if (base64 == null) {
+            return McpResponses.error(
+                    mapper, "Missing required argument: file (base64-encoded content).");
+        }
+        byte[] bytes = McpToolSupport.decodeBase64OrNull(base64);
+        if (bytes == null) {
+            return McpResponses.error(mapper, "The 'file' argument is not valid base64.");
+        }
+        String name = McpToolSupport.textArg(arguments, "fileName");
+        if (name == null) {
+            name = "upload.bin";
+        }
+        try {
+            String fileId = fileStorage.storeBytes(bytes, name);
+            return McpResponses.text(
+                    mapper,
+                    "Stored '"
+                            + name
+                            + "' ("
+                            + bytes.length
+                            + " bytes) as fileId="
+                            + fileId
+                            + ". Pass this fileId to a Stirling operation's 'fileId' argument.");
+        } catch (IOException e) {
+            log.warn("MCP upload failed to store file", e);
+            return McpResponses.error(mapper, "Failed to store the uploaded file.");
+        }
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/model/api/ai/AiWorkflowOutcome.java

@@ -21,7 +21,8 @@ public enum AiWorkflowOutcome {
     COMPLETED("completed"),
     UNSUPPORTED_CAPABILITY("unsupported_capability"),
     CANNOT_CONTINUE("cannot_continue"),
-    GENERATE_FILE("generate_file");
+    GENERATE_FILE("generate_file"),
+    CONVERT_MARKDOWN("convert_markdown");
 
     private final String value;
 

app/proprietary/src/main/java/stirling/software/proprietary/policy/config/FolderAccessGuard.java

@@ -0,0 +1,106 @@
+package stirling.software.proprietary.policy.config;
+
+import java.nio.file.Path;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+
+import org.springframework.core.env.Environment;
+import org.springframework.stereotype.Component;
+
+import stirling.software.common.configuration.InstallationPathConfig;
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.proprietary.policy.model.Policy;
+
+/**
+ * The single authority on which filesystem locations a policy may read from or write to. Folder
+ * sources and sinks take a configured directory, so without this a user who can save a policy could
+ * point one at Stirling's own config/secrets directory and exfiltrate (or overwrite) it. Every
+ * folder source and sink runs its directory through {@link #requirePermitted(Path)} at save time
+ * and again at run time.
+ *
+ * <p>Enforced fail-closed, in order:
+ *
+ * <ul>
+ *   <li><b>Disabled in SaaS</b> - folder access is never allowed when the {@code saas} profile is
+ *       active; a tenant must not reach the host filesystem at all.
+ *   <li><b>Protected paths</b> - Stirling's own config directory (settings, database, keys,
+ *       backups) is always rejected, even if an allowed root were misconfigured to contain it.
+ *   <li><b>Allowlist</b> - the directory must resolve within one of {@code
+ *       policies.allowedFolderRoots}; with none configured, all folder access is refused.
+ * </ul>
+ *
+ * <p>Paths are compared after normalisation, so {@code ..} segments cannot walk out of an allowed
+ * root. (Symlink escape is not defended here; an operator who configures an allowed root containing
+ * a symlink to a sensitive location is trusted.)
+ */
+@Component
+public class FolderAccessGuard {
+
+    public static final String FOLDER_TYPE = "folder";
+
+    private final boolean saasActive;
+    private final List<Path> allowedRoots;
+    private final List<Path> protectedRoots;
+
+    public FolderAccessGuard(ApplicationProperties applicationProperties, Environment environment) {
+        this.saasActive = Arrays.asList(environment.getActiveProfiles()).contains("saas");
+        this.allowedRoots =
+                normalizeAll(applicationProperties.getPolicies().getAllowedFolderRoots());
+        this.protectedRoots = List.of(normalize(Path.of(InstallationPathConfig.getConfigPath())));
+    }
+
+    /**
+     * Check that {@code dir} is a permitted folder location, returning its normalised absolute
+     * form.
+     *
+     * @throws IllegalArgumentException if folder access is disabled (SaaS or no roots configured),
+     *     the path is inside a protected directory, or it falls outside every allowed root
+     */
+    public Path requirePermitted(Path dir) {
+        if (saasActive) {
+            throw new IllegalArgumentException(
+                    "folder sources and outputs are not available in SaaS mode");
+        }
+        Path normalized = normalize(dir);
+        for (Path protectedRoot : protectedRoots) {
+            if (normalized.startsWith(protectedRoot)) {
+                throw new IllegalArgumentException(
+                        "folder may not point inside a protected Stirling directory");
+            }
+        }
+        if (allowedRoots.isEmpty()) {
+            throw new IllegalArgumentException(
+                    "folder access is disabled; set policies.allowedFolderRoots to permit it");
+        }
+        boolean within = allowedRoots.stream().anyMatch(normalized::startsWith);
+        if (!within) {
+            throw new IllegalArgumentException(
+                    "folder '" + normalized + "' is outside the allowed folder roots");
+        }
+        return normalized;
+    }
+
+    /** Whether this policy reads from or writes to a folder, and so is subject to these rules. */
+    public boolean usesFolderAccess(Policy policy) {
+        boolean readsFolder =
+                policy.sources().stream().anyMatch(spec -> FOLDER_TYPE.equals(spec.type()));
+        boolean writesFolder =
+                policy.output() != null && FOLDER_TYPE.equals(policy.output().type());
+        return readsFolder || writesFolder;
+    }
+
+    private static List<Path> normalizeAll(List<String> roots) {
+        List<Path> result = new ArrayList<>();
+        for (String root : roots) {
+            if (root != null && !root.isBlank()) {
+                result.add(normalize(Path.of(root)));
+            }
+        }
+        return result;
+    }
+
+    private static Path normalize(Path path) {
+        return path.toAbsolutePath().normalize();
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/controller/PolicyController.java

@@ -0,0 +1,351 @@
+package stirling.software.proprietary.policy.controller;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.LinkedHashMap;
+import java.util.List;
+import java.util.Map;
+
+import org.springframework.core.io.FileSystemResource;
+import org.springframework.core.io.Resource;
+import org.springframework.http.HttpStatus;
+import org.springframework.http.MediaType;
+import org.springframework.http.ResponseEntity;
+import org.springframework.util.MultiValueMap;
+import org.springframework.web.bind.annotation.DeleteMapping;
+import org.springframework.web.bind.annotation.GetMapping;
+import org.springframework.web.bind.annotation.PathVariable;
+import org.springframework.web.bind.annotation.PostMapping;
+import org.springframework.web.bind.annotation.RequestBody;
+import org.springframework.web.bind.annotation.RequestMapping;
+import org.springframework.web.bind.annotation.RequestParam;
+import org.springframework.web.bind.annotation.RestController;
+import org.springframework.web.multipart.MultipartFile;
+import org.springframework.web.multipart.MultipartHttpServletRequest;
+import org.springframework.web.server.ResponseStatusException;
+import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
+
+import io.github.pixee.security.Filenames;
+import io.swagger.v3.oas.annotations.Hidden;
+import io.swagger.v3.oas.annotations.Operation;
+import io.swagger.v3.oas.annotations.tags.Tag;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.model.job.JobResponse;
+import stirling.software.common.service.UserServiceInterface;
+import stirling.software.common.util.TempFile;
+import stirling.software.common.util.TempFileManager;
+import stirling.software.proprietary.policy.config.FolderAccessGuard;
+import stirling.software.proprietary.policy.engine.PolicyRunHandle;
+import stirling.software.proprietary.policy.engine.PolicyRunRegistry;
+import stirling.software.proprietary.policy.engine.PolicyRunner;
+import stirling.software.proprietary.policy.engine.PolicyValidator;
+import stirling.software.proprietary.policy.model.PipelineDefinition;
+import stirling.software.proprietary.policy.model.Policy;
+import stirling.software.proprietary.policy.model.PolicyInputs;
+import stirling.software.proprietary.policy.model.PolicyRun;
+import stirling.software.proprietary.policy.model.PolicyRunStatus;
+import stirling.software.proprietary.policy.model.PolicyRunView;
+import stirling.software.proprietary.policy.progress.PolicyProgressListener;
+import stirling.software.proprietary.policy.store.PolicyStore;
+import stirling.software.proprietary.security.config.PremiumEndpoint;
+
+import tools.jackson.core.JacksonException;
+import tools.jackson.databind.ObjectMapper;
+
+/**
+ * Manages policies and runs pipelines. The premium backend entry point: CRUD for stored {@code
+ * Policy} objects, running a stored policy by id, and running an ad-hoc pipeline (for AI/Automate
+ * one-offs).
+ *
+ * <p>Runs execute asynchronously and return a run id immediately. Poll {@code GET /run/{runId}} for
+ * status, and download outputs via the existing {@code GET /api/v1/general/files/{fileId}} using
+ * the file ids in the run view.
+ */
+@Slf4j
+@RestController
+@RequestMapping("/api/v1/policies")
+@Hidden
+@PremiumEndpoint
+@RequiredArgsConstructor
+@Tag(name = "Policies", description = "Run tool pipelines on the backend")
+public class PolicyController {
+
+    private final PolicyRunner policyRunner;
+    private final PolicyRunRegistry runRegistry;
+    private final PolicyStore policyStore;
+    private final PolicyValidator policyValidator;
+    private final FolderAccessGuard folderAccessGuard;
+    private final UserServiceInterface userService;
+    private final ApplicationProperties applicationProperties;
+    private final ObjectMapper objectMapper;
+    private final TempFileManager tempFileManager;
+
+    @PostMapping(value = "/run", consumes = MediaType.MULTIPART_FORM_DATA_VALUE)
+    @Operation(
+            summary = "Run a tool pipeline",
+            description =
+                    "Accepts the documents to process (multipart field 'fileInput'), any supporting"
+                            + " files (each under a multipart field named as its asset key, e.g."
+                            + " 'company-logo'), and a JSON pipeline definition ('json'). Runs the"
+                            + " steps in order asynchronously and returns a run id. Poll the run"
+                            + " status endpoint and download outputs via /api/v1/general/files/{id}.")
+    public ResponseEntity<JobResponse<Void>> run(
+            @RequestParam("json") String json, MultipartHttpServletRequest request)
+            throws IOException {
+        PipelineDefinition definition = parseDefinition(json);
+        PolicyInputs inputs = collectInputs(request);
+        String runId =
+                policyRunner.runAdHoc(definition, inputs, PolicyProgressListener.NOOP).runId();
+        return ResponseEntity.accepted().body(new JobResponse<>(true, runId, null));
+    }
+
+    @PostMapping(value = "/run/stream", consumes = MediaType.MULTIPART_FORM_DATA_VALUE)
+    @Operation(
+            summary = "Run a tool pipeline with live progress",
+            description =
+                    "Same as /run, but returns Server-Sent Events: a 'step' event as each step"
+                            + " starts and completes, then a terminal 'completed', 'failed',"
+                            + " 'cancelled', or 'waiting' event carrying the final run view.")
+    public SseEmitter runStream(
+            @RequestParam("json") String json, MultipartHttpServletRequest request)
+            throws IOException {
+        PipelineDefinition definition = parseDefinition(json);
+        PolicyInputs inputs = collectInputs(request);
+
+        SseEmitter emitter =
+                new SseEmitter(applicationProperties.getPolicies().getStreamTimeoutMs());
+        emitter.onError(e -> log.warn("Policy run SSE emitter error", e));
+
+        PolicyRunHandle handle = policyRunner.runAdHoc(definition, inputs, streamListener(emitter));
+        // Close the stream with a terminal event once the run finishes. whenComplete runs on the
+        // engine's worker thread after the run is done, so this never races the step events.
+        handle.completion()
+                .whenComplete(
+                        (run, throwable) -> {
+                            if (throwable != null) {
+                                sendEvent(
+                                        emitter,
+                                        "failed",
+                                        Map.of("message", throwable.getMessage()));
+                            } else {
+                                sendEvent(emitter, terminalEventName(run), PolicyRunView.of(run));
+                            }
+                            emitter.complete();
+                        });
+        return emitter;
+    }
+
+    @GetMapping("/run/{runId}")
+    @Operation(
+            summary = "Get pipeline run status",
+            description = "Returns the current status, step cursor, and output files of a run.")
+    public ResponseEntity<PolicyRunView> status(@PathVariable String runId) {
+        PolicyRun run = runRegistry.get(runId);
+        if (run == null) {
+            return ResponseEntity.notFound().build();
+        }
+        return ResponseEntity.ok(PolicyRunView.of(run));
+    }
+
+    // --- Policy management ---
+
+    @PostMapping(consumes = MediaType.APPLICATION_JSON_VALUE)
+    @Operation(
+            summary = "Create or update a policy",
+            description =
+                    "Stores a policy (trigger config + steps + output + metadata). A blank id is"
+                            + " assigned; returns the stored policy with its id.")
+    public ResponseEntity<Policy> savePolicy(@RequestBody String json) {
+        Policy policy = parsePolicy(json);
+        requireAuthorizedForFolderAccess(policy);
+        try {
+            policyValidator.validate(policy);
+        } catch (IllegalArgumentException e) {
+            throw new ResponseStatusException(HttpStatus.BAD_REQUEST, e.getMessage());
+        }
+        return ResponseEntity.ok(policyStore.save(policy));
+    }
+
+    /**
+     * A policy that reads from or writes to a server folder grants whoever saves it access to that
+     * path, so restrict it to administrators on multi-user deployments. Single-user deployments
+     * (login disabled, e.g. desktop) trust the local operator. The {@link FolderAccessGuard} still
+     * enforces SaaS-off and the path allowlist during validation regardless of who saves.
+     */
+    private void requireAuthorizedForFolderAccess(Policy policy) {
+        if (!folderAccessGuard.usesFolderAccess(policy)) {
+            return;
+        }
+        if (!applicationProperties.getSecurity().isEnableLogin()) {
+            return;
+        }
+        if (!userService.isCurrentUserAdmin()) {
+            throw new ResponseStatusException(
+                    HttpStatus.FORBIDDEN,
+                    "Folder sources and outputs may only be configured by an administrator");
+        }
+    }
+
+    @GetMapping
+    @Operation(summary = "List policies")
+    public List<Policy> listPolicies() {
+        return policyStore.all();
+    }
+
+    @GetMapping("/{policyId}")
+    @Operation(summary = "Get a policy by id")
+    public ResponseEntity<Policy> getPolicy(@PathVariable String policyId) {
+        return policyStore
+                .get(policyId)
+                .map(ResponseEntity::ok)
+                .orElseGet(() -> ResponseEntity.notFound().build());
+    }
+
+    @DeleteMapping("/{policyId}")
+    @Operation(summary = "Delete a policy by id")
+    public ResponseEntity<Void> deletePolicy(@PathVariable String policyId) {
+        return policyStore.delete(policyId)
+                ? ResponseEntity.noContent().build()
+                : ResponseEntity.notFound().build();
+    }
+
+    @PostMapping(value = "/{policyId}/run", consumes = MediaType.MULTIPART_FORM_DATA_VALUE)
+    @Operation(
+            summary = "Run a stored policy",
+            description =
+                    "Runs the stored policy's pipeline on the supplied files (primary documents"
+                            + " under 'fileInput', supporting files under their asset-key fields)."
+                            + " Runs regardless of the policy's enabled flag, which only gates"
+                            + " automatic triggering. Returns a run id.")
+    public ResponseEntity<JobResponse<Void>> runStoredPolicy(
+            @PathVariable String policyId, MultipartHttpServletRequest request) throws IOException {
+        Policy policy =
+                policyStore
+                        .get(policyId)
+                        .orElseThrow(
+                                () ->
+                                        new ResponseStatusException(
+                                                HttpStatus.NOT_FOUND, "No policy: " + policyId));
+        PolicyInputs inputs = collectInputs(request);
+        String runId = policyRunner.runWith(policy, inputs, PolicyProgressListener.NOOP).runId();
+        return ResponseEntity.accepted().body(new JobResponse<>(true, runId, null));
+    }
+
+    private Policy parsePolicy(String json) {
+        try {
+            return objectMapper.readValue(json, Policy.class);
+        } catch (JacksonException e) {
+            throw new ResponseStatusException(HttpStatus.BAD_REQUEST, "Invalid policy JSON");
+        }
+    }
+
+    private PipelineDefinition parseDefinition(String json) {
+        PipelineDefinition definition;
+        try {
+            definition = objectMapper.readValue(json, PipelineDefinition.class);
+        } catch (JacksonException e) {
+            throw new ResponseStatusException(
+                    HttpStatus.BAD_REQUEST, "Invalid pipeline definition JSON");
+        }
+        if (definition.steps().isEmpty()) {
+            throw new ResponseStatusException(
+                    HttpStatus.BAD_REQUEST, "Pipeline definition has no steps");
+        }
+        return definition;
+    }
+
+    /**
+     * Split the multipart file parts into the primary document stream ("fileInput") and the named
+     * supporting-file store: every other file field becomes an asset keyed by its field name, which
+     * a step references from {@code fileParameters}.
+     */
+    private PolicyInputs collectInputs(MultipartHttpServletRequest request) throws IOException {
+        MultiValueMap<String, MultipartFile> fileMap = request.getMultiFileMap();
+        List<Resource> primary = toResources(fileMap.get("fileInput"));
+        Map<String, List<Resource>> supportingFiles = new LinkedHashMap<>();
+        for (Map.Entry<String, List<MultipartFile>> entry : fileMap.entrySet()) {
+            if ("fileInput".equals(entry.getKey())) {
+                continue;
+            }
+            List<Resource> assets = toResources(entry.getValue());
+            if (!assets.isEmpty()) {
+                supportingFiles.put(entry.getKey(), assets);
+            }
+        }
+        return new PolicyInputs(primary, supportingFiles);
+    }
+
+    /**
+     * A progress listener that forwards each step transition to the SSE stream as a "step" event.
+     */
+    private PolicyProgressListener streamListener(SseEmitter emitter) {
+        return new PolicyProgressListener() {
+            @Override
+            public void onStepStart(int stepIndex, int stepCount, String operation) {
+                sendEvent(emitter, "step", stepEvent("started", stepIndex, stepCount, operation));
+            }
+
+            @Override
+            public void onStepComplete(int stepIndex, int stepCount, String operation) {
+                sendEvent(emitter, "step", stepEvent("completed", stepIndex, stepCount, operation));
+            }
+        };
+    }
+
+    private static Map<String, Object> stepEvent(
+            String phase, int stepIndex, int stepCount, String operation) {
+        return Map.of(
+                "phase", phase,
+                "stepIndex", stepIndex,
+                "stepCount", stepCount,
+                "operation", operation);
+    }
+
+    private static String terminalEventName(PolicyRun run) {
+        PolicyRunStatus status = run.getStatus();
+        return switch (status) {
+            case COMPLETED -> "completed";
+            case FAILED -> "failed";
+            case CANCELLED -> "cancelled";
+            case WAITING_FOR_INPUT -> "waiting";
+            default -> "ended";
+        };
+    }
+
+    private void sendEvent(SseEmitter emitter, String name, Object data) {
+        try {
+            emitter.send(SseEmitter.event().name(name).data(data, MediaType.APPLICATION_JSON));
+        } catch (IOException | IllegalStateException e) {
+            // Client disconnected or the emitter already closed. The run continues and its results
+            // remain downloadable via the job endpoints; nothing useful left to stream.
+            log.debug("Dropping policy SSE event '{}': {}", name, e.getMessage());
+        }
+    }
+
+    private List<Resource> toResources(List<MultipartFile> files) throws IOException {
+        List<Resource> resources = new ArrayList<>();
+        if (files == null) {
+            return resources;
+        }
+        for (MultipartFile file : files) {
+            if (file == null || file.isEmpty()) {
+                continue;
+            }
+            TempFile tempFile = tempFileManager.createManagedTempFile("policy-run");
+            file.transferTo(tempFile.getPath());
+            final String originalName = Filenames.toSimpleFileName(file.getOriginalFilename());
+            resources.add(
+                    new FileSystemResource(tempFile.getFile()) {
+                        @Override
+                        public String getFilename() {
+                            return originalName;
+                        }
+                    });
+        }
+        return resources;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyEngine.java

@@ -0,0 +1,261 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.UUID;
+import java.util.concurrent.CompletableFuture;
+import java.util.concurrent.ExecutorService;
+
+import org.springframework.core.io.Resource;
+import org.springframework.http.ResponseEntity;
+import org.springframework.stereotype.Service;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.job.ResultFile;
+import stirling.software.common.service.FileStorage;
+import stirling.software.common.service.InternalApiTimeoutException;
+import stirling.software.common.service.JobOwnershipService;
+import stirling.software.common.service.JobQueue;
+import stirling.software.common.service.ResourceMonitor;
+import stirling.software.common.service.TaskManager;
+import stirling.software.common.util.ExecutorFactory;
+import stirling.software.proprietary.policy.model.OutputSpec;
+import stirling.software.proprietary.policy.model.PipelineDefinition;
+import stirling.software.proprietary.policy.model.Policy;
+import stirling.software.proprietary.policy.model.PolicyInputs;
+import stirling.software.proprietary.policy.model.PolicyRun;
+import stirling.software.proprietary.policy.model.WaitState;
+import stirling.software.proprietary.policy.output.PolicyOutputSink;
+import stirling.software.proprietary.policy.progress.PolicyProgressListener;
+
+/**
+ * Runs pipelines asynchronously as tracked jobs.
+ *
+ * <p>Each run is the unit of async work: {@link #submit} returns a run id immediately and the
+ * pipeline executes on a virtual thread, so a step blocking on a slow tool does not tie up a
+ * platform thread. The run drives {@link PolicyExecutor} for the actual step loop, registers its
+ * outputs and progress with {@link TaskManager} (so the existing job status/download endpoints work
+ * unchanged), and keeps rich state in {@link PolicyRunRegistry}.
+ *
+ * <p>The engine deliberately manages its own virtual-thread execution rather than routing through
+ * {@code JobExecutorService}: that path force-completes a job once its work returns, which is
+ * incompatible with a run that suspends in {@code WAITING_FOR_INPUT}. It still applies the shared
+ * {@link ResourceMonitor}/{@link JobQueue} admission control, so heavy runs queue under load
+ * instead of oversubscribing.
+ */
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class PolicyEngine {
+
+    /**
+     * Resource weight of a pipeline run for admission control. A run chains many tools and holds
+     * intermediate files, so it is weighted as heavy work: the shared {@link ResourceMonitor}
+     * should let it start while the system is healthy but hold it back under memory/CPU pressure.
+     * See {@link ResourceMonitor#shouldQueueJob(int)} for how a weight maps to that decision.
+     */
+    private static final int RUN_RESOURCE_WEIGHT = 50;
+
+    private final PolicyExecutor stepExecutor;
+    private final TaskManager taskManager;
+    private final PolicyRunRegistry registry;
+    private final FileStorage fileStorage;
+    private final JobOwnershipService jobOwnershipService;
+    private final List<PolicyOutputSink> outputSinks;
+    private final ResourceMonitor resourceMonitor;
+    private final JobQueue jobQueue;
+
+    private final ExecutorService asyncExecutor = ExecutorFactory.newVirtualThreadExecutor();
+
+    /**
+     * Submit a pipeline to run asynchronously. The returned handle's run id scopes a job in {@link
+     * TaskManager}, so progress (notes), status, and result files are observable via the existing
+     * job endpoints as well as via {@link #getRun(String)}; its completion future resolves when the
+     * run reaches a terminal or paused state.
+     */
+    public PolicyRunHandle submit(
+            PipelineDefinition definition, PolicyInputs inputs, PolicyProgressListener listener) {
+        // Scope the run id to the current user (on this request thread) so the file-download
+        // ownership check passes; NoOpJobOwnershipService returns the id unchanged when security
+        // is off.
+        String runId = jobOwnershipService.createScopedJobKey(UUID.randomUUID().toString());
+        taskManager.createTask(runId);
+        PolicyRun run = new PolicyRun(runId, definition);
+        registry.register(run);
+        CompletableFuture<PolicyRun> completion = new CompletableFuture<>();
+        PolicyProgressListener tracking = trackingListener(runId, run, listener);
+        Runnable task = () -> runToCompletion(run, inputs, tracking, completion);
+
+        // Each run is one admission unit; steps run synchronously within it, so this gates heavy
+        // work under load without the pool-within-pool risk of queueing each tool call. Under
+        // resource pressure the run waits in the shared JobQueue; otherwise it starts immediately.
+        if (resourceMonitor.shouldQueueJob(RUN_RESOURCE_WEIGHT)) {
+            log.debug("Queueing policy run {} under resource pressure", runId);
+            jobQueue.queueJob(
+                            runId,
+                            RUN_RESOURCE_WEIGHT,
+                            () -> {
+                                task.run();
+                                return null;
+                            },
+                            0L)
+                    .exceptionally(ex -> failRejectedRun(run, completion, ex));
+        } else {
+            asyncExecutor.execute(task);
+        }
+        return new PolicyRunHandle(runId, completion);
+    }
+
+    /**
+     * Run a stored policy on demand. Builds the policy's pipeline and submits it. {@code enabled}
+     * gates automatic triggering, not explicit runs, so this runs regardless of that flag.
+     */
+    public PolicyRunHandle runPolicy(
+            Policy policy, PolicyInputs inputs, PolicyProgressListener listener) {
+        return submit(policy.toDefinition(), inputs, listener);
+    }
+
+    public PolicyRun getRun(String runId) {
+        return registry.get(runId);
+    }
+
+    /**
+     * Request cancellation of a run. Stage 1 marks the run cancelled in the registry if it has not
+     * already finished; interrupting an in-flight tool call lands in a later stage.
+     */
+    public boolean cancel(String runId) {
+        PolicyRun run = registry.get(runId);
+        if (run == null) {
+            return false;
+        }
+        boolean cancelled = run.cancel();
+        if (cancelled) {
+            taskManager.addNote(runId, "Run cancelled by request");
+        }
+        return cancelled;
+    }
+
+    /**
+     * Resume a run paused in {@code WAITING_FOR_INPUT}. Not yet implemented; the run shape and
+     * {@link WaitState} snapshot are in place so this can be added without reworking the engine.
+     */
+    public String resume(String runId, List<Resource> additionalInputs) {
+        throw new UnsupportedOperationException("Pause/resume is not yet implemented");
+    }
+
+    private void runToCompletion(
+            PolicyRun run,
+            PolicyInputs inputs,
+            PolicyProgressListener listener,
+            CompletableFuture<PolicyRun> completion) {
+        String runId = run.getRunId();
+        try {
+            run.markRunning();
+            PolicyExecutionResult result =
+                    stepExecutor.execute(run.getDefinition(), inputs, listener);
+            OutputSpec output = run.getDefinition().output();
+            List<ResultFile> outputs = sinkFor(output).deliver(runId, result.files(), output);
+            taskManager.setMultipleFileResults(runId, outputs);
+            taskManager.setComplete(runId);
+            run.complete(outputs);
+        } catch (PolicyInputRequiredException e) {
+            // Designed-for path: suspend the run rather than fail it. Persist intermediates as
+            // fileIds so the run can resume after this worker thread is gone.
+            WaitState wait = suspend(e);
+            run.waitForInput(wait);
+            taskManager.addNote(runId, "Waiting for input: " + e.getMessage());
+        } catch (InternalApiTimeoutException e) {
+            String message = toolTimeoutMessage(e);
+            log.error(
+                    "Policy run {} timed out on {}: {}",
+                    runId,
+                    e.getEndpointPath(),
+                    e.getMessage());
+            run.fail(message);
+            taskManager.setError(runId, message);
+        } catch (Exception e) {
+            String message = "Policy run failed: " + e.getMessage();
+            log.error("Policy run {} failed", runId, e);
+            run.fail(message);
+            taskManager.setError(runId, message);
+        } finally {
+            // Always resolve the handle with the run's final state so stream/await callers unblock.
+            completion.complete(run);
+        }
+    }
+
+    private ResponseEntity<?> failRejectedRun(
+            PolicyRun run, CompletableFuture<PolicyRun> completion, Throwable ex) {
+        // Only reached if the run never started (e.g. the queue was full). A run that started
+        // always resolves its own completion in runToCompletion.
+        if (!completion.isDone()) {
+            String message = "Policy run could not be queued: " + ex.getMessage();
+            log.error("Policy run {} was not admitted: {}", run.getRunId(), ex.getMessage());
+            run.fail(message);
+            taskManager.setError(run.getRunId(), message);
+            completion.complete(run);
+        }
+        return null;
+    }
+
+    private WaitState suspend(PolicyInputRequiredException e) {
+        List<String> fileIds = new ArrayList<>();
+        for (Resource resource : e.getPendingFiles()) {
+            String name = resource.getFilename() != null ? resource.getFilename() : "pending";
+            try (InputStream is = resource.getInputStream()) {
+                fileIds.add(fileStorage.storeInputStream(is, name).fileId());
+            } catch (IOException io) {
+                log.warn("Failed to persist pending file for paused run: {}", io.getMessage());
+            }
+        }
+        return new WaitState(e.getMessage(), e.getResumeStepIndex(), fileIds);
+    }
+
+    private PolicyProgressListener trackingListener(
+            String runId, PolicyRun run, PolicyProgressListener delegate) {
+        return new PolicyProgressListener() {
+            @Override
+            public void onStepStart(int stepIndex, int stepCount, String operation) {
+                run.enterStep(stepIndex);
+                taskManager.addNote(
+                        runId,
+                        "Step " + stepIndex + "/" + stepCount + ": " + operation + " started");
+                delegate.onStepStart(stepIndex, stepCount, operation);
+            }
+
+            @Override
+            public void onStepComplete(int stepIndex, int stepCount, String operation) {
+                taskManager.addNote(
+                        runId,
+                        "Step " + stepIndex + "/" + stepCount + ": " + operation + " completed");
+                delegate.onStepComplete(stepIndex, stepCount, operation);
+            }
+
+            @Override
+            public void onHeartbeat() {
+                delegate.onHeartbeat();
+            }
+        };
+    }
+
+    private PolicyOutputSink sinkFor(OutputSpec spec) {
+        return outputSinks.stream()
+                .filter(sink -> sink.supports(spec))
+                .findFirst()
+                .orElseThrow(
+                        () ->
+                                new IllegalStateException(
+                                        "No output sink supports spec: "
+                                                + (spec == null ? "<null>" : spec.type())));
+    }
+
+    private static String toolTimeoutMessage(InternalApiTimeoutException e) {
+        return String.format(
+                "The %s tool did not respond within %d seconds and was aborted.",
+                e.getEndpointPath(), e.getReadTimeout().toSeconds());
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyExecutionResult.java

@@ -0,0 +1,17 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.util.List;
+
+import org.springframework.core.io.Resource;
+
+import tools.jackson.databind.JsonNode;
+
+/**
+ * Result of running a pipeline through {@link PolicyExecutor}.
+ *
+ * <p>{@code files} are the final output resources (temp files, not yet stored to {@code
+ * FileStorage}). {@code report} is the structured metadata payload captured from the last step that
+ * produced one (a JSON body, or an {@code X-Stirling-Tool-Report} header), with {@code reportTool}
+ * naming the step it came from; both are null when no step produced a report.
+ */
+public record PolicyExecutionResult(List<Resource> files, JsonNode report, String reportTool) {}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyExecutor.java

@@ -0,0 +1,312 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Locale;
+import java.util.Map;
+
+import org.springframework.core.io.Resource;
+import org.springframework.http.HttpHeaders;
+import org.springframework.http.HttpStatus;
+import org.springframework.http.MediaType;
+import org.springframework.http.ResponseEntity;
+import org.springframework.stereotype.Service;
+import org.springframework.util.LinkedMultiValueMap;
+import org.springframework.util.MultiValueMap;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.service.InternalApiClient;
+import stirling.software.common.service.InternalApiTimeoutException;
+import stirling.software.common.service.ToolMetadataService;
+import stirling.software.common.util.TempFileManager;
+import stirling.software.common.util.ZipExtractionUtils;
+import stirling.software.proprietary.policy.model.PipelineDefinition;
+import stirling.software.proprietary.policy.model.PipelineStep;
+import stirling.software.proprietary.policy.model.PolicyInputs;
+import stirling.software.proprietary.policy.progress.PolicyProgressListener;
+import stirling.software.proprietary.service.AiToolResponseHeaders;
+
+import tools.jackson.core.JacksonException;
+import tools.jackson.databind.JsonNode;
+import tools.jackson.databind.ObjectMapper;
+
+/**
+ * Runs an ordered chain of tool steps, chaining each step's output files into the next step's
+ * input.
+ *
+ * <p>This is the single execution loop for the proprietary surface (AI plans now;
+ * manually-triggered runs and watched folders later). Each step is dispatched synchronously via
+ * {@link InternalApiClient} loopback HTTP: the tool runs in its own handler and returns its file
+ * inline. The caller decides how to run the executor itself (the AI turn loop calls it directly;
+ * the engine runs it on a virtual thread for async runs). Files cross step boundaries as {@link
+ * Resource} temp files; they are only persisted to durable storage at the run boundaries by the
+ * caller.
+ */
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class PolicyExecutor {
+
+    private static final String FILTER_OPERATION_PREFIX = "/api/v1/filter/filter-";
+
+    private final InternalApiClient internalApiClient;
+    private final ToolMetadataService toolMetadataService;
+    private final TempFileManager tempFileManager;
+    private final ObjectMapper objectMapper;
+
+    /**
+     * Internal value-class for tool responses. {@code files} holds any result files (typically one;
+     * multiple for ZIP-response tools). {@code report} holds an optional structured metadata
+     * payload the tool chose to surface alongside (or instead of) a file.
+     */
+    private record ToolResult(List<Resource> files, JsonNode report) {}
+
+    /**
+     * Execute every step in {@code definition} in order, feeding each step's output into the next.
+     * Supporting files supplied in {@code inputs} are bound to steps' named file fields and never
+     * enter the document stream.
+     *
+     * @param definition the pipeline to run (must have at least one step)
+     * @param inputs the primary documents plus the named supporting-file store
+     * @param listener receives per-step progress
+     * @return the final output files plus the last structured report produced, if any
+     * @throws InternalApiTimeoutException if a tool does not respond within its read timeout
+     * @throws IOException if a tool returns a non-OK response, references a missing supporting
+     *     file, or a file cannot be read
+     */
+    public PolicyExecutionResult execute(
+            PipelineDefinition definition, PolicyInputs inputs, PolicyProgressListener listener)
+            throws IOException {
+        List<PipelineStep> steps = definition.steps();
+        if (steps.isEmpty()) {
+            throw new IllegalArgumentException("Pipeline definition has no steps");
+        }
+
+        List<Resource> currentFiles = inputs.primary();
+        Map<String, List<Resource>> supportingFiles = inputs.supportingFiles();
+        // Propagate the *last* non-null report; the terminal step defines the output.
+        JsonNode lastReport = null;
+        String lastReportTool = null;
+
+        for (int i = 0; i < steps.size(); i++) {
+            PipelineStep step = steps.get(i);
+            String operation = step.operation();
+            if (operation == null || operation.isBlank()) {
+                throw new IllegalArgumentException(
+                        "Pipeline step " + (i + 1) + " has no operation");
+            }
+            listener.onStepStart(i + 1, steps.size(), operation);
+            ToolResult stepResult = executeStep(step, currentFiles, supportingFiles);
+            currentFiles = stepResult.files();
+            if (stepResult.report() != null) {
+                lastReport = stepResult.report();
+                lastReportTool = operation;
+            }
+            listener.onStepComplete(i + 1, steps.size(), operation);
+        }
+
+        return new PolicyExecutionResult(currentFiles, lastReport, lastReportTool);
+    }
+
+    /**
+     * Execute a single tool step. If the endpoint accepts multiple files, all files are sent in one
+     * call. Otherwise, the endpoint is called once per file. ZIP responses are unpacked so each
+     * inner file is treated as its own result (e.g. split outputs a ZIP of pages).
+     *
+     * <p>A structured {@code report} may be returned alongside (or instead of) files; see {@link
+     * ToolResult}. For per-file dispatch (single-input endpoints called once per input), the first
+     * non-null report wins.
+     */
+    private ToolResult executeStep(
+            PipelineStep step,
+            List<Resource> inputFiles,
+            Map<String, List<Resource>> supportingFiles)
+            throws IOException {
+        requireAcceptedTypes(step.operation(), inputFiles);
+        List<Resource> files = new ArrayList<>();
+        JsonNode report = null;
+        if (toolMetadataService.isMultiInput(step.operation())) {
+            ToolResult r = callEndpoint(step, inputFiles, supportingFiles);
+            files.addAll(r.files());
+            report = r.report();
+        } else {
+            for (Resource file : inputFiles) {
+                ToolResult r = callEndpoint(step, List.of(file), supportingFiles);
+                files.addAll(r.files());
+                if (report == null) {
+                    report = r.report();
+                }
+            }
+        }
+        return new ToolResult(files, report);
+    }
+
+    /**
+     * Call an endpoint and return its result files and optional report.
+     *
+     * <ul>
+     *   <li>JSON body (Content-Type: application/json): the entire body is the report, no files are
+     *       returned.
+     *   <li>File body (PDF etc.): the file is returned; if an {@link
+     *       AiToolResponseHeaders#TOOL_REPORT} header is present, its (minified JSON) value is
+     *       parsed as the report.
+     *   <li>ZIP responses declared by the tool metadata service are unpacked so callers always see
+     *       a flat list of result files.
+     * </ul>
+     */
+    private ToolResult callEndpoint(
+            PipelineStep step, List<Resource> files, Map<String, List<Resource>> supportingFiles)
+            throws IOException {
+        String endpointPath = step.operation();
+        MultiValueMap<String, Object> body = new LinkedMultiValueMap<>();
+        for (Resource file : files) {
+            body.add("fileInput", file);
+        }
+        // Bind supporting files to their named tool fields (e.g. stampImage, overlayFiles). These
+        // come from the run's named asset store, not the document stream.
+        for (Map.Entry<String, String> binding : step.fileParameters().entrySet()) {
+            String fieldName = binding.getKey();
+            String assetKey = binding.getValue();
+            List<Resource> assets = supportingFiles.get(assetKey);
+            if (assets == null || assets.isEmpty()) {
+                throw new IOException(
+                        "Step "
+                                + endpointPath
+                                + " references supporting file '"
+                                + assetKey
+                                + "' for field '"
+                                + fieldName
+                                + "' but no such file was provided");
+            }
+            for (Resource asset : assets) {
+                body.add(fieldName, asset);
+            }
+        }
+        for (Map.Entry<String, Object> entry : step.parameters().entrySet()) {
+            if (entry.getValue() instanceof List<?> list) {
+                if (containsStructuredElements(list)) {
+                    // Endpoints binding lists of structured objects (e.g. /security/redact's
+                    // redactions, /general/edit-text's edits) parse a single JSON string field via
+                    // a property editor. Pre-serialize the whole list so binding succeeds.
+                    body.add(entry.getKey(), objectMapper.writeValueAsString(list));
+                } else {
+                    for (Object item : list) {
+                        body.add(entry.getKey(), item);
+                    }
+                }
+            } else {
+                body.add(entry.getKey(), entry.getValue());
+            }
+        }
+        ResponseEntity<Resource> response = internalApiClient.post(endpointPath, body);
+        if (!HttpStatus.OK.equals(response.getStatusCode()) || response.getBody() == null) {
+            throw new IOException(
+                    "Tool returned HTTP " + response.getStatusCode() + " for " + endpointPath);
+        }
+        Resource resource = response.getBody();
+
+        // Filter operations return an empty body to signal the file was filtered out: drop it
+        // rather than forwarding a zero-byte document.
+        if (isFilterOperation(endpointPath) && isEmpty(resource)) {
+            return new ToolResult(List.of(), null);
+        }
+
+        HttpHeaders headers = response.getHeaders();
+        MediaType contentType = headers.getContentType();
+
+        // JSON-only response: the whole body is the structured report, no result file.
+        if (contentType != null && MediaType.APPLICATION_JSON.isCompatibleWith(contentType)) {
+            try (InputStream is = resource.getInputStream()) {
+                JsonNode report = objectMapper.readTree(is);
+                return new ToolResult(List.of(), report);
+            }
+        }
+
+        JsonNode report = parseReportHeader(headers, endpointPath);
+        if (toolMetadataService.shouldUnpackZipResponse(endpointPath)) {
+            return new ToolResult(ZipExtractionUtils.extractZip(resource, tempFileManager), report);
+        }
+        return new ToolResult(List.of(resource), report);
+    }
+
+    /**
+     * Parse the optional {@link AiToolResponseHeaders#TOOL_REPORT} header into a {@link JsonNode},
+     * or return null.
+     */
+    private JsonNode parseReportHeader(HttpHeaders headers, String endpointPath) {
+        String raw = headers.getFirst(AiToolResponseHeaders.TOOL_REPORT);
+        if (raw == null || raw.isBlank()) {
+            return null;
+        }
+        try {
+            return objectMapper.readTree(raw);
+        } catch (JacksonException e) {
+            log.warn(
+                    "Ignoring malformed {} header from {}: {}",
+                    AiToolResponseHeaders.TOOL_REPORT,
+                    endpointPath,
+                    e.getMessage());
+            return null;
+        }
+    }
+
+    private static boolean containsStructuredElements(List<?> list) {
+        for (Object item : list) {
+            if (item instanceof Map<?, ?> || item instanceof List<?>) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+    /**
+     * Fail the run if any document in the primary stream is not a file type the step accepts. An
+     * endpoint that declares no specific input type accepts anything.
+     */
+    private void requireAcceptedTypes(String operation, List<Resource> files) throws IOException {
+        List<String> accepted = toolMetadataService.getExtensionTypes(false, operation);
+        if (accepted == null || accepted.isEmpty()) {
+            return;
+        }
+        for (Resource file : files) {
+            if (!matchesType(file, accepted)) {
+                throw new IOException(
+                        "Step "
+                                + operation
+                                + " accepts "
+                                + accepted
+                                + " but received '"
+                                + file.getFilename()
+                                + "'");
+            }
+        }
+    }
+
+    private static boolean matchesType(Resource file, List<String> acceptedExtensions) {
+        String filename = file.getFilename();
+        if (filename == null) {
+            return false;
+        }
+        int dot = filename.lastIndexOf('.');
+        if (dot < 0 || dot == filename.length() - 1) {
+            return false;
+        }
+        return acceptedExtensions.contains(filename.substring(dot + 1).toLowerCase(Locale.ROOT));
+    }
+
+    private static boolean isFilterOperation(String operation) {
+        return operation.startsWith(FILTER_OPERATION_PREFIX);
+    }
+
+    private static boolean isEmpty(Resource resource) {
+        try {
+            return resource.contentLength() == 0;
+        } catch (IOException e) {
+            return false;
+        }
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyInputRequiredException.java

@@ -0,0 +1,32 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.util.List;
+
+import org.springframework.core.io.Resource;
+
+import lombok.Getter;
+
+/**
+ * Thrown by a step to signal that the run cannot proceed without further user input, pausing the
+ * run in {@code WAITING_FOR_INPUT} rather than failing it.
+ *
+ * <p>Carries everything needed to resume: a human-readable reason, the 0-based index of the step to
+ * resume from, and the intermediate files produced so far. The engine persists those files and
+ * suspends the run.
+ *
+ * <p>Defined now to fix the run shape; no step throws it yet, and the resume handshake is
+ * implemented in a later stage.
+ */
+@Getter
+public class PolicyInputRequiredException extends RuntimeException {
+
+    private final transient List<Resource> pendingFiles;
+    private final int resumeStepIndex;
+
+    public PolicyInputRequiredException(
+            String reason, int resumeStepIndex, List<Resource> pendingFiles) {
+        super(reason);
+        this.resumeStepIndex = resumeStepIndex;
+        this.pendingFiles = pendingFiles == null ? List.of() : pendingFiles;
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyRunHandle.java

@@ -0,0 +1,16 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.util.concurrent.CompletableFuture;
+
+import stirling.software.proprietary.policy.model.PolicyRun;
+
+/**
+ * Returned by {@link PolicyEngine#submit}: the run id (for status polling and result download) plus
+ * a future that resolves when the run reaches a terminal or paused state.
+ *
+ * <p>The completion future lets callers react to the end of a run (e.g. an SSE endpoint sending a
+ * final event and closing the stream) without polling. It carries the {@link PolicyRun} whose
+ * status describes the outcome (completed, failed, cancelled, or waiting for input); it does not
+ * complete exceptionally for ordinary run failures.
+ */
+public record PolicyRunHandle(String runId, CompletableFuture<PolicyRun> completion) {}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyRunRegistry.java

@@ -0,0 +1,97 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.time.Duration;
+import java.time.Instant;
+import java.util.Collection;
+import java.util.Map;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.concurrent.Executors;
+import java.util.concurrent.ScheduledExecutorService;
+import java.util.concurrent.TimeUnit;
+
+import org.springframework.stereotype.Service;
+
+import jakarta.annotation.PreDestroy;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.proprietary.policy.model.PolicyRun;
+
+/**
+ * In-memory store of live {@link PolicyRun} state, keyed by runId. Holds the authoritative run
+ * state machine; durable status/files for download are projected separately into {@code
+ * TaskManager}.
+ *
+ * <p>Finished runs are evicted on a fixed interval once they age past {@code
+ * policies.runExpiryMinutes}, mirroring the job-result expiry in {@code TaskManager} so a run's
+ * rich in-memory state does not outlive the process. Only terminal runs are evicted; active and
+ * paused ({@code WAITING_FOR_INPUT}) runs are retained regardless of age. Result files are not
+ * touched here: a run shares its runId with a {@code TaskManager} job, which owns file-lifecycle
+ * cleanup, so eviction only frees this map's entry.
+ */
+@Slf4j
+@Service
+public class PolicyRunRegistry {
+
+    private final Map<String, PolicyRun> runs = new ConcurrentHashMap<>();
+
+    private final Duration runExpiry;
+    private final ScheduledExecutorService cleanupExecutor =
+            Executors.newSingleThreadScheduledExecutor(
+                    Thread.ofVirtual().name("policy-run-cleanup-", 0).factory());
+
+    public PolicyRunRegistry(ApplicationProperties applicationProperties) {
+        int runExpiryMinutes = applicationProperties.getPolicies().getRunExpiryMinutes();
+        this.runExpiry = Duration.ofMinutes(runExpiryMinutes);
+        cleanupExecutor.scheduleAtFixedRate(this::evictExpiredRuns, 10, 10, TimeUnit.MINUTES);
+        log.debug(
+                "Policy run registry initialized with run expiry of {} minutes", runExpiryMinutes);
+    }
+
+    public void register(PolicyRun run) {
+        runs.put(run.getRunId(), run);
+    }
+
+    public PolicyRun get(String runId) {
+        return runs.get(runId);
+    }
+
+    public Collection<PolicyRun> all() {
+        return runs.values();
+    }
+
+    /** Scheduled hook: evict terminal runs that finished before the expiry window. */
+    private void evictExpiredRuns() {
+        try {
+            evictExpired(Instant.now().minus(runExpiry));
+        } catch (Exception e) {
+            log.error("Error during policy run cleanup: {}", e.getMessage(), e);
+        }
+    }
+
+    /**
+     * Remove every terminal run last updated before {@code cutoff}; active and paused runs are kept
+     * regardless of age. Returns the number evicted. Package-visible so the scheduled sweep and
+     * tests exercise the same path with an explicit cutoff.
+     */
+    int evictExpired(Instant cutoff) {
+        int removed = 0;
+        for (Map.Entry<String, PolicyRun> entry : runs.entrySet()) {
+            PolicyRun run = entry.getValue();
+            if (run.getStatus().isTerminal() && run.getUpdatedAt().isBefore(cutoff)) {
+                runs.remove(entry.getKey());
+                removed++;
+            }
+        }
+        if (removed > 0) {
+            log.info("Evicted {} expired policy runs", removed);
+        }
+        return removed;
+    }
+
+    @PreDestroy
+    void shutdown() {
+        cleanupExecutor.shutdownNow();
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyRunner.java

@@ -0,0 +1,116 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.io.IOException;
+import java.util.List;
+import java.util.function.Consumer;
+
+import org.springframework.stereotype.Service;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.proprietary.policy.input.InputSource;
+import stirling.software.proprietary.policy.input.ResolvedInput;
+import stirling.software.proprietary.policy.model.InputSpec;
+import stirling.software.proprietary.policy.model.PipelineDefinition;
+import stirling.software.proprietary.policy.model.Policy;
+import stirling.software.proprietary.policy.model.PolicyInputs;
+import stirling.software.proprietary.policy.model.PolicyRun;
+import stirling.software.proprietary.policy.model.PolicyRunStatus;
+import stirling.software.proprietary.policy.progress.PolicyProgressListener;
+
+/**
+ * Runs policies, and is the one place that knows how to turn a policy's configured {@link InputSpec
+ * sources} into actual runs. Triggers (schedule, and future webhook/folder-watch) decide
+ * <em>when</em> to run and call {@link #run(Policy)}; they never touch sources themselves. The
+ * controller uses the supplied-input and ad-hoc entry points for on-demand work.
+ *
+ * <p>This is the seam that keeps triggers and sources independent: a trigger depends on the runner,
+ * the runner depends on the {@link InputSource} beans, and a source depends on neither - it just
+ * yields {@link ResolvedInput units of work}, each carrying its own completion hook.
+ */
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class PolicyRunner {
+
+    private final PolicyEngine policyEngine;
+    private final List<InputSource> inputSources;
+
+    /**
+     * Run a policy by pulling from every source it configures: each source yields zero or more
+     * units of work, and each unit becomes its own run so one failure does not affect the others. A
+     * policy with no sources runs once with no input files (a generator pipeline). Used by
+     * automatic triggers.
+     */
+    public void run(Policy policy) {
+        List<InputSpec> sources = policy.sources();
+        if (sources.isEmpty()) {
+            startRun(policy, PolicyInputs.of(List.of()), unused -> {});
+            return;
+        }
+        for (InputSpec spec : sources) {
+            pullAndRun(policy, spec);
+        }
+    }
+
+    /**
+     * Run a stored policy on files supplied directly by the caller (e.g. a manual run with
+     * uploads), bypassing its configured sources. Returns the run handle so callers can stream
+     * progress.
+     */
+    public PolicyRunHandle runWith(
+            Policy policy, PolicyInputs inputs, PolicyProgressListener listener) {
+        return policyEngine.runPolicy(policy, inputs, listener);
+    }
+
+    /** Run an ad-hoc pipeline with no stored policy (AI/Automate one-offs). */
+    public PolicyRunHandle runAdHoc(
+            PipelineDefinition definition, PolicyInputs inputs, PolicyProgressListener listener) {
+        return policyEngine.submit(definition, inputs, listener);
+    }
+
+    private void pullAndRun(Policy policy, InputSpec spec) {
+        InputSource source = sourceFor(spec);
+        if (source == null) {
+            log.warn(
+                    "No input source for type '{}' (policy {}); skipping",
+                    spec.type(),
+                    policy.id());
+            return;
+        }
+        List<ResolvedInput> work;
+        try {
+            work = source.resolve(spec);
+        } catch (IOException | RuntimeException e) {
+            log.warn(
+                    "Failed to resolve source '{}' for policy {}: {}",
+                    spec.type(),
+                    policy.id(),
+                    e.getMessage());
+            return;
+        }
+        for (ResolvedInput unit : work) {
+            startRun(policy, unit.inputs(), unit.onComplete());
+        }
+    }
+
+    private void startRun(Policy policy, PolicyInputs inputs, Consumer<Boolean> onComplete) {
+        log.info("Running policy {} ({})", policy.id(), policy.name());
+        PolicyRunHandle handle =
+                policyEngine.runPolicy(policy, inputs, PolicyProgressListener.NOOP);
+        handle.completion()
+                .whenComplete((run, throwable) -> onComplete.accept(succeeded(run, throwable)));
+    }
+
+    private static boolean succeeded(PolicyRun run, Throwable throwable) {
+        return throwable == null && run != null && run.getStatus() == PolicyRunStatus.COMPLETED;
+    }
+
+    private InputSource sourceFor(InputSpec spec) {
+        return inputSources.stream()
+                .filter(source -> source.supports(spec))
+                .findFirst()
+                .orElse(null);
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/engine/PolicyValidator.java

@@ -0,0 +1,74 @@
+package stirling.software.proprietary.policy.engine;
+
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+
+import lombok.RequiredArgsConstructor;
+
+import stirling.software.proprietary.policy.input.InputSource;
+import stirling.software.proprietary.policy.model.InputSpec;
+import stirling.software.proprietary.policy.model.OutputSpec;
+import stirling.software.proprietary.policy.model.Policy;
+import stirling.software.proprietary.policy.model.TriggerConfig;
+import stirling.software.proprietary.policy.output.PolicyOutputSink;
+import stirling.software.proprietary.policy.trigger.PolicyTrigger;
+
+/**
+ * Validates a policy's trigger, sources, and output configuration by delegating each facet to the
+ * bean that handles its type. Called when a policy is saved so a misconfigured schedule, missing
+ * folder directory, or unknown type fails fast instead of silently misbehaving at run time.
+ *
+ * <p>The trigger is optional (a {@code null} trigger is a manual-only policy and needs no
+ * validation); every configured source is validated.
+ */
+@Service
+@RequiredArgsConstructor
+public class PolicyValidator {
+
+    private final List<PolicyTrigger> triggers;
+    private final List<InputSource> inputSources;
+    private final List<PolicyOutputSink> outputSinks;
+
+    /**
+     * @throws IllegalArgumentException if any facet's type is unknown or its configuration is
+     *     invalid
+     */
+    public void validate(Policy policy) {
+        if (policy.trigger() != null) {
+            triggerFor(policy.trigger()).validate(policy);
+        }
+        for (InputSpec source : policy.sources()) {
+            inputSourceFor(source).validate(source);
+        }
+        outputSinkFor(policy.output()).validate(policy.output());
+    }
+
+    private PolicyTrigger triggerFor(TriggerConfig config) {
+        return triggers.stream()
+                .filter(trigger -> trigger.type().equals(config.type()))
+                .findFirst()
+                .orElseThrow(
+                        () ->
+                                new IllegalArgumentException(
+                                        "unknown trigger type: " + config.type()));
+    }
+
+    private InputSource inputSourceFor(InputSpec spec) {
+        return inputSources.stream()
+                .filter(source -> source.supports(spec))
+                .findFirst()
+                .orElseThrow(
+                        () ->
+                                new IllegalArgumentException(
+                                        "unknown input source type: " + spec.type()));
+    }
+
+    private PolicyOutputSink outputSinkFor(OutputSpec spec) {
+        return outputSinks.stream()
+                .filter(sink -> sink.supports(spec))
+                .findFirst()
+                .orElseThrow(
+                        () -> new IllegalArgumentException("unknown output type: " + spec.type()));
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/input/FolderInputSource.java

@@ -0,0 +1,186 @@
+package stirling.software.proprietary.policy.input;
+
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.StandardCopyOption;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Map;
+import java.util.stream.Stream;
+
+import org.springframework.core.io.FileSystemResource;
+import org.springframework.core.io.Resource;
+import org.springframework.stereotype.Service;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.util.FileReadinessChecker;
+import stirling.software.proprietary.policy.config.FolderAccessGuard;
+import stirling.software.proprietary.policy.model.InputSpec;
+import stirling.software.proprietary.policy.model.PolicyInputs;
+
+/**
+ * Reads input files from a directory. Each ready file becomes its own unit of work (one run per
+ * file) so a failure on one file does not affect the others.
+ *
+ * <p>Two modes via the {@code mode} option:
+ *
+ * <ul>
+ *   <li>{@code "consume"} (default) - claim each file by moving it into {@code
+ *       .stirling/processing}, then route it to {@code .stirling/done} or {@code .stirling/error}
+ *       when its run finishes. Each file is processed once; right for "process new arrivals" (and
+ *       the basis of watched folders).
+ *   <li>{@code "snapshot"} - read the directory's current files without moving them; every run sees
+ *       the full set again. Right for "always regenerate from a fixed input set".
+ * </ul>
+ *
+ * Readiness is checked first (via {@link FileReadinessChecker}) so files mid-write are skipped.
+ */
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class FolderInputSource implements InputSource {
+
+    private static final String TYPE = FolderAccessGuard.FOLDER_TYPE;
+    // Bookkeeping lives under one hidden namespace dir so the watched folder stays tidy.
+    private static final String WORK_SUBDIR = ".stirling";
+    private static final String PROCESSING_SUBDIR = "processing";
+    private static final String DONE_SUBDIR = "done";
+    private static final String ERROR_SUBDIR = "error";
+
+    private final FileReadinessChecker readinessChecker;
+    private final FolderAccessGuard accessGuard;
+
+    @Override
+    public String type() {
+        return TYPE;
+    }
+
+    @Override
+    public boolean supports(InputSpec spec) {
+        return spec != null && TYPE.equals(spec.type());
+    }
+
+    @Override
+    public void validate(InputSpec spec) {
+        accessGuard.requirePermitted(FolderConfig.from(spec.options()).directory());
+    }
+
+    @Override
+    public List<Path> watchTargets(InputSpec spec) {
+        return List.of(FolderConfig.from(spec.options()).directory());
+    }
+
+    @Override
+    public List<ResolvedInput> resolve(InputSpec spec) throws IOException {
+        FolderConfig config = FolderConfig.from(spec.options());
+        Path inputDir = accessGuard.requirePermitted(config.directory());
+        if (!Files.isDirectory(inputDir)) {
+            log.debug("Folder input dir does not exist: {}", inputDir);
+            return List.of();
+        }
+
+        List<Path> ready = new ArrayList<>();
+        try (Stream<Path> entries = Files.list(inputDir)) {
+            entries.filter(Files::isRegularFile)
+                    .filter(readinessChecker::isReady)
+                    .forEach(ready::add);
+        }
+
+        List<ResolvedInput> work = new ArrayList<>();
+        for (Path file : ready) {
+            if (config.snapshot()) {
+                work.add(ResolvedInput.of(PolicyInputs.of(List.of(fileResource(file)))));
+            } else {
+                Path claimed = claim(inputDir, file);
+                if (claimed == null) {
+                    continue; // another sweep/process grabbed it
+                }
+                work.add(
+                        new ResolvedInput(
+                                PolicyInputs.of(List.of(fileResource(claimed))),
+                                success -> route(inputDir, claimed, success)));
+            }
+        }
+        return work;
+    }
+
+    private Path claim(Path inputDir, Path file) {
+        try {
+            Path processingDir = workDir(inputDir, PROCESSING_SUBDIR);
+            Files.createDirectories(processingDir);
+            Path claimed = uniqueTarget(processingDir, file.getFileName().toString());
+            Files.move(file, claimed, StandardCopyOption.ATOMIC_MOVE);
+            return claimed;
+        } catch (IOException e) {
+            log.debug("Could not claim {}: {}", file, e.getMessage());
+            return null;
+        }
+    }
+
+    private void route(Path inputDir, Path claimed, boolean success) {
+        String subdir = success ? DONE_SUBDIR : ERROR_SUBDIR;
+        try {
+            Path destDir = workDir(inputDir, subdir);
+            Files.createDirectories(destDir);
+            Files.move(
+                    claimed,
+                    uniqueTarget(destDir, claimed.getFileName().toString()),
+                    StandardCopyOption.ATOMIC_MOVE);
+        } catch (IOException e) {
+            log.warn(
+                    "Could not move processed input {} to {}: {}", claimed, subdir, e.getMessage());
+        }
+    }
+
+    /** A bookkeeping subdirectory under the watched folder's {@code .stirling} namespace. */
+    private static Path workDir(Path inputDir, String subdir) {
+        return inputDir.resolve(WORK_SUBDIR).resolve(subdir);
+    }
+
+    private static Resource fileResource(Path path) {
+        String name = path.getFileName().toString();
+        return new FileSystemResource(path.toFile()) {
+            @Override
+            public String getFilename() {
+                return name;
+            }
+        };
+    }
+
+    private static Path uniqueTarget(Path dir, String filename) {
+        Path candidate = dir.resolve(filename);
+        if (!Files.exists(candidate)) {
+            return candidate;
+        }
+        int dot = filename.lastIndexOf('.');
+        String base = dot < 0 ? filename : filename.substring(0, dot);
+        String ext = dot < 0 ? "" : filename.substring(dot);
+        for (int n = 1; ; n++) {
+            Path next = dir.resolve(base + " (" + n + ")" + ext);
+            if (!Files.exists(next)) {
+                return next;
+            }
+        }
+    }
+
+    /** The typed, validated form of a folder source's options: the directory and dedup mode. */
+    record FolderConfig(Path directory, boolean snapshot) {
+
+        private static final String DIRECTORY_OPTION = "directory";
+        private static final String MODE_OPTION = "mode";
+        private static final String MODE_SNAPSHOT = "snapshot";
+
+        static FolderConfig from(Map<String, Object> options) {
+            Object directory = options.get(DIRECTORY_OPTION);
+            if (directory == null || directory.toString().isBlank()) {
+                throw new IllegalArgumentException("folder input requires a 'directory' option");
+            }
+            Object mode = options.get(MODE_OPTION);
+            boolean snapshot = mode != null && MODE_SNAPSHOT.equals(mode.toString());
+            return new FolderConfig(Path.of(directory.toString()), snapshot);
+        }
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/input/InputSource.java

@@ -0,0 +1,49 @@
+package stirling.software.proprietary.policy.input;
+
+import java.io.IOException;
+import java.nio.file.Path;
+import java.util.List;
+
+import stirling.software.proprietary.policy.model.InputSpec;
+
+/**
+ * Resolves one of a policy's {@link InputSpec sources} into the files to run on - answering
+ * <em>where</em> a run's files come from, independent of <em>when</em> it runs. The counterpart of
+ * {@code PolicyOutputSink}: implementations are beans selected by {@link #supports(InputSpec)}, so
+ * a new source kind (folder, S3) is just a new bean.
+ *
+ * <p>Driven by the {@code PolicyRunner}, which a trigger calls when a policy is due; a source is
+ * passive and knows nothing about what triggered the run. A manual run may instead supply files
+ * directly and bypass sources entirely.
+ */
+public interface InputSource {
+
+    /** Stable identifier for this source, matching {@code InputSpec.type()} (e.g. "folder"). */
+    String type();
+
+    /** Whether this source can handle the given spec. */
+    boolean supports(InputSpec spec);
+
+    /**
+     * Check that an input spec is usable, throwing {@link IllegalArgumentException} if not. Called
+     * when a policy is saved so misconfiguration fails fast rather than at run time.
+     */
+    default void validate(InputSpec spec) {}
+
+    /**
+     * Resolve the spec into zero or more units of work, each carrying the files for one run and a
+     * completion hook. Returning an empty list means there is nothing to run right now.
+     */
+    List<ResolvedInput> resolve(InputSpec spec) throws IOException;
+
+    /**
+     * The local filesystem directories this source draws from, if any, for triggers that want to
+     * react to changes there (the folder-watch trigger) rather than poll. Advisory only: it merely
+     * tells a trigger <em>where</em> to watch; resolving the spec into files is still this source's
+     * job via {@link #resolve}. Non-filesystem sources (S3, ...) return an empty list and so are
+     * simply not watchable. Default: nothing to watch.
+     */
+    default List<Path> watchTargets(InputSpec spec) {
+        return List.of();
+    }
+}

app/proprietary/src/main/java/stirling/software/proprietary/policy/input/ResolvedInput.java

@@ -0,0 +1,23 @@
+package stirling.software.proprietary.policy.input;
+
+import java.util.function.Consumer;
+
+import stirling.software.proprietary.policy.model.PolicyInputs;
+
+/**
+ * One unit of work produced by an {@link InputSource}: the files to run plus a completion callback
+ * invoked with the run's success once it finishes (e.g. a folder source routes the input to {@code
+ * .stirling/done} or {@code .stirling/error}). A source may return several of these (e.g. one per
+ * file).
+ */
+public record ResolvedInput(PolicyInputs inputs, Consumer<Boolean> onComplete) {
+
+    public ResolvedInput {
+        onComplete = onComplete == null ? success -> {} : onComplete;
+    }
+
+    /** A unit of work with no completion side effect. */
+    public static ResolvedInput of(PolicyInputs inputs) {
+        return new ResolvedInput(inputs, success -> {});
+    }
+}