fix(security): scope cert-sign participant listing to caller's team (GHSA-h2rx-xrhc-5q72)

Signed-off-by: Balázs Szücs <bszucs1209@gmail.com>

.dockerignore

@@ -26,8 +26,9 @@ version_builds/
 node_modules/
 **/node_modules/
 frontend/node_modules/
-frontend/dist/
-frontend/playwright-report/
+frontend/editor/dist/
+frontend/dist-portal/
+frontend/editor/playwright-report/
 .npm/
 .yarn/
 

.github/actions/setup-bot/action.yml

@@ -24,7 +24,7 @@ runs:
       id: generate-token
       uses: actions/create-github-app-token@bcd2ba49218906704ab6c1aa796996da409d3eb1 # v3.2.0
       with:
-        app-id: ${{ inputs.app-id }}
+        client-id: ${{ inputs.app-id }}
         private-key: ${{ inputs.private-key }}
     - name: Configure Git
       run: |

.github/aur/stirling-pdf-desktop/PKGBUILD

@@ -1,6 +1,6 @@
 # Maintainer: Stirling PDF Inc <contact@stirlingpdf.com>
 pkgname=stirling-pdf-desktop
-pkgver=2.11.0
+pkgver=2.12.0
 pkgrel=1
 pkgdesc="Locally hosted, web-based PDF manipulation tool (Tauri desktop app, official Stirling PDF Inc build)"
 arch=('x86_64')

.github/aur/stirling-pdf-server-bin/PKGBUILD

@@ -1,6 +1,6 @@
 # Maintainer: Stirling PDF Inc <contact@stirlingpdf.com>
 pkgname=stirling-pdf-server-bin
-pkgver=2.11.0
+pkgver=2.12.0
 pkgrel=1
 pkgdesc="Locally hosted, web-based PDF manipulation tool (server JAR, prebuilt)"
 arch=('any')

.github/config/.files.yaml

@@ -38,6 +38,8 @@ project: &project
   - frontend/**
   - docker/**
   - scripts/RestartHelper.java
+  - scripts/db-migration/**
+  - .github/workflows/db-migration-test.yml
 
 frontend: &frontend
   - frontend/**

.github/scripts/check_language_toml.py

@@ -13,7 +13,7 @@ Usage:
 """
 
 # Sample for Windows:
-# python .github/scripts/check_language_toml.py --reference-file frontend/public/locales/en-GB/translation.toml --branch "" --files frontend/public/locales/de-DE/translation.toml frontend/public/locales/fr-FR/translation.toml
+# python .github/scripts/check_language_toml.py --reference-file frontend/editor/public/locales/en-GB/translation.toml --branch "" --files frontend/editor/public/locales/de-DE/translation.toml frontend/editor/public/locales/fr-FR/translation.toml
 
 import argparse
 import glob
@@ -184,7 +184,7 @@ def check_for_differences(reference_file, file_list, branch, actor):
     if len(file_list) == 1:
         file_arr = file_list[0].split()
 
-    base_dir = Path.cwd() / "frontend" / "public" / "locales"
+    base_dir = Path.cwd() / "frontend" / "editor" / "public" / "locales"
 
     for file_path in file_arr:
         file_path = Path(file_path)
@@ -308,7 +308,7 @@ def check_for_differences(reference_file, file_list, branch, actor):
         report.append("## ❌ Overall Check Status: **_Failed_**")
         report.append("")
         report.append(
-            f"@{actor} please check your translation if it conforms to the standard. Follow the format of [en-GB/translation.toml](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/frontend/public/locales/en-GB/translation.toml)"
+            f"@{actor} please check your translation if it conforms to the standard. Follow the format of [en-GB/translation.toml](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/frontend/editor/public/locales/en-GB/translation.toml)"
         )
     else:
         report.append("## ✅ Overall Check Status: **_Success_**")
@@ -372,6 +372,7 @@ if __name__ == "__main__":
                 os.path.join(
                     os.getcwd(),
                     "frontend",
+                    "editor",
                     "public",
                     "locales",
                     "*",

.github/scripts/verify-updater-signatures.py

@@ -0,0 +1,83 @@
+#!/usr/bin/env python3
+"""Verify Tauri updater .sig files against plugins.updater.pubkey in tauri.conf.json.
+
+Usage: verify-updater-signatures.py <dir-to-scan> [tauri.conf.json]
+"""
+
+import binascii
+import sys
+import json
+import base64
+import hashlib
+from pathlib import Path
+from cryptography.hazmat.primitives.asymmetric.ed25519 import Ed25519PublicKey
+from cryptography.exceptions import InvalidSignature
+
+ART_ROOT = Path(sys.argv[1])
+CONF = Path(
+    sys.argv[2] if len(sys.argv) > 2 else "frontend/editor/src-tauri/tauri.conf.json"
+)
+
+
+def load_pubkey():
+    # tauri pubkey = base64 of a minisign .pub file; last line is base64 of
+    # [2 algo][8 key-id][32 ed25519 public key].
+    raw = json.loads(CONF.read_text())["plugins"]["updater"]["pubkey"]
+    blob = base64.b64decode(base64.b64decode(raw).decode().splitlines()[-1])
+    return blob[2:10], Ed25519PublicKey.from_public_bytes(blob[10:])
+
+
+def hash_file(path: Path) -> bytes:
+    h = hashlib.blake2b(digest_size=64)
+    with path.open("rb") as f:
+        for chunk in iter(lambda: f.read(1 << 16), b""):
+            h.update(chunk)
+    return h.digest()
+
+
+def verify(artifact: Path, sig_file: Path, keyid_pub, pub) -> str:
+    # tauri .sig = base64 of a minisign signature file (4 lines).
+    try:
+        lines = base64.b64decode(sig_file.read_text()).decode().splitlines()
+        sig_blob = base64.b64decode(lines[1])
+    except (binascii.Error, IndexError, UnicodeDecodeError) as e:
+        return f"FAIL malformed sig ({type(e).__name__})"
+    algo, keyid, sig = sig_blob[:2], sig_blob[2:10], sig_blob[10:74]
+    if keyid != keyid_pub:
+        return f"FAIL key-id mismatch (sig {keyid.hex()} vs pub {keyid_pub.hex()})"
+    # 'ED' = prehashed (BLAKE2b-512), 'Ed' = legacy (raw message).
+    msg = hash_file(artifact) if algo == b"ED" else artifact.read_bytes()
+    try:
+        pub.verify(sig, msg)
+    except InvalidSignature:
+        return f"FAIL signature invalid (algo={algo.decode()})"
+    # Global signature covers sig + trusted_comment.
+    gc = "global-sig FAIL"
+    try:
+        tc = lines[2].split("trusted comment: ", 1)[1]
+        pub.verify(base64.b64decode(lines[3]), sig + tc.encode())
+        gc = "global-sig OK"
+    except (InvalidSignature, IndexError, binascii.Error):
+        pass
+    return f"VALID (algo={algo.decode()}, keyid={keyid.hex()}, {gc})"
+
+
+keyid_pub, pub = load_pubkey()
+print(f"updater pubkey keyid={keyid_pub.hex()}\n")
+sigs = sorted(ART_ROOT.rglob("*.sig"))
+if not sigs:
+    print(f"WARN: no .sig files under {ART_ROOT} - nothing to verify")
+    sys.exit(0)
+bad = 0
+for sig_file in sigs:
+    artifact = sig_file.with_suffix("")
+    if not artifact.exists():
+        print(f"  ? {sig_file.name}: artifact missing")
+        bad += 1
+        continue
+    res = verify(artifact, sig_file, keyid_pub, pub)
+    print(f"  {artifact.name}: {res}")
+    if not res.startswith("VALID") or "global-sig FAIL" in res:
+        bad += 1
+print(f"\n{'ALL SIGNATURES VALID' if bad == 0 else f'{bad} SIGNATURE(S) FAILED'}")
+sys.exit(1 if bad else 0)

.github/workflows/PR-Auto-Deploy-V2.yml

@@ -287,6 +287,7 @@ jobs:
                 - /stirling/V2-PR-${{ needs.check-pr.outputs.pr_number }}/data:/usr/share/tessdata:rw
                 - /stirling/V2-PR-${{ needs.check-pr.outputs.pr_number }}/config:/configs:rw
                 - /stirling/V2-PR-${{ needs.check-pr.outputs.pr_number }}/logs:/logs:rw
+                - /stirling/V2-PR-${{ needs.check-pr.outputs.pr_number }}/storage:/storage:rw
               environment:
                 DISABLE_ADDITIONAL_FEATURES: "false"
                 SECURITY_ENABLELOGIN: "true"
@@ -309,7 +310,7 @@ jobs:
 
           ssh -i ../private.key -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -T ${{ secrets.NEW_VPS_USERNAME }}@${{ secrets.NEW_VPS_HOST }} << ENDSSH
             # Create V2 PR-specific directories
-            mkdir -p /stirling/V2-PR-${{ needs.check-pr.outputs.pr_number }}/{data,config,logs}
+            mkdir -p /stirling/V2-PR-${{ needs.check-pr.outputs.pr_number }}/{data,config,logs,storage}
 
             # Move docker-compose file to correct location
             mv /tmp/docker-compose-v2.yml /stirling/V2-PR-${{ needs.check-pr.outputs.pr_number }}/docker-compose.yml

.github/workflows/ai-engine.yml

@@ -1,8 +1,9 @@
 name: AI Engine CI
 
-# Validates the Python AI engine: regenerates tool models, runs fixers,
-# lint, type-check, and tests. Called from build.yml on PRs and merge_group;
-# also runs directly on push to main as a post-merge safety net.
+# Validates the Python AI engine: regenerates tool models and runs the
+# engine quality gate (lint, type-check, format-check, tests). Called from
+# build.yml on PRs and merge_group; also runs directly on push to main as
+# a post-merge safety net.
 on:
   workflow_call:
   push:
@@ -51,27 +52,95 @@ jobs:
         run: task engine:tool-models
 
       - name: Verify tool models are up to date
+        id: tool-models-check
+        continue-on-error: true
+        run: git diff --exit-code engine/src/stirling/models/tool_models.py
+
+      - name: Comment on tool models check failure
+        # Only post a comment on PRs. github-script's PR helpers need an
+        # issue/PR number, which doesn't exist on merge_group runs.
+        if: steps.tool-models-check.outcome == 'failure' && github.event_name == 'pull_request'
+        continue-on-error: true
+        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
+        with:
+          script: |
+            const marker = '<!-- tool-models-check -->';
+            const body = [
+              marker,
+              '### Tool Models Check Failed',
+              '',
+              'The generated `engine/src/stirling/models/tool_models.py` is out of date with the Java OpenAPI spec and will need to be regenerated before it can be merged in.',
+              '',
+              'Run `task engine:tool-models` to regenerate, then commit the updated file.',
+            ].join('\n');
+            const { data: comments } = await github.rest.issues.listComments({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: context.issue.number,
+            });
+            const existing = comments.find(c => c.body.includes(marker));
+            if (existing) {
+              await github.rest.issues.updateComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                comment_id: existing.id,
+                body,
+              });
+            } else {
+              await github.rest.issues.createComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                issue_number: context.issue.number,
+                body,
+              });
+            }
+
+      - name: Fail if tool models check failed
+        if: steps.tool-models-check.outcome == 'failure'
         run: |
-          if ! git diff --exit-code engine/src/stirling/models/tool_models.py; then
-            echo "tool_models.py is out of date."
-            echo "Run 'task engine:tool-models' locally and commit the updated file."
-            exit 1
-          fi
+          echo "============================================"
+          echo "  Tool Models Check Failed"
+          echo "============================================"
+          echo ""
+          echo "The generated engine/src/stirling/models/tool_models.py"
+          echo "is out of date with the Java OpenAPI spec and will"
+          echo "need to be regenerated before it can be merged in."
+          echo ""
+          echo "Run 'task engine:tool-models' to regenerate, then"
+          echo "commit the updated file."
+          echo "============================================"
+          exit 1
 
-      - name: Run fixers
-        run: task engine:fix
+      - name: Remove tool models check comment on success
+        if: steps.tool-models-check.outcome == 'success' && github.event_name == 'pull_request'
+        continue-on-error: true
+        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
+        with:
+          script: |
+            const marker = '<!-- tool-models-check -->';
+            const { data: comments } = await github.rest.issues.listComments({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: context.issue.number,
+            });
+            const existing = comments.find(c => c.body.includes(marker));
+            if (existing) {
+              await github.rest.issues.deleteComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                comment_id: existing.id,
+              });
+            }
 
-      - name: Verify fixes are committed
-        id: fixer_changes
-        run: |
-          if ! git diff --quiet; then
-            git --no-pager diff --stat
-            echo "::error::There are issues with your Python code that will need to be fixed before they can be merged in. Run 'task engine:fix' to auto-fix what can be fixed automatically, then run 'task engine:check' to see what still needs fixing manually."
-            exit 1
-          fi
+      - name: Quality-check engine
+        id: engine-check
+        run: task engine:check
+        continue-on-error: true
 
-      - name: Comment on fixer failures
-        if: steps.fixer_changes.outcome == 'failure' && github.event_name == 'pull_request'
+      - name: Comment on engine check failure
+        # Only post a comment on PRs. github-script's PR helpers need an
+        # issue/PR number, which doesn't exist on merge_group runs.
+        if: steps.engine-check.outcome == 'failure' && github.event_name == 'pull_request'
         continue-on-error: true
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         with:
@@ -107,11 +176,39 @@ jobs:
               });
             }
 
-      - name: Run linting
-        run: task engine:lint
+      - name: Fail if engine check failed
+        if: steps.engine-check.outcome == 'failure'
+        run: |
+          echo "============================================"
+          echo "  Engine Check Failed"
+          echo "============================================"
+          echo ""
+          echo "There are issues with your Python code that"
+          echo "will need to be fixed before they can be merged in."
+          echo ""
+          echo "Run 'task engine:fix' to auto-fix what can be"
+          echo "fixed automatically, then run 'task engine:check'"
+          echo "to see what still needs fixing manually."
+          echo "============================================"
+          exit 1
 
-      - name: Run type checking
-        run: task engine:typecheck
-
-      - name: Run tests
-        run: task engine:test
+      - name: Remove engine check comment on success
+        if: steps.engine-check.outcome == 'success' && github.event_name == 'pull_request'
+        continue-on-error: true
+        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
+        with:
+          script: |
+            const marker = '<!-- engine-check -->';
+            const { data: comments } = await github.rest.issues.listComments({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: context.issue.number,
+            });
+            const existing = comments.find(c => c.body.includes(marker));
+            if (existing) {
+              await github.rest.issues.deleteComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                comment_id: existing.id,
+              });
+            }

.github/workflows/backend-build.yml

@@ -1,8 +1,14 @@
 name: Backend build, format check, and coverage
 
 # Reusable workflow called from build.yml. Runs the backend build matrix
-# (JDK 25 × spring-security on/off), Spotless formatting check, JUnit, and
+# (JDK 25 × every flavor), Spotless formatting check, JUnit, and
 # posts Jacoco coverage to PRs.
+#
+# Flavor axis (maps to STIRLING_FLAVOR in settings.gradle):
+#   core         - DISABLE_ADDITIONAL_FEATURES=true, no proprietary, no saas
+#   proprietary  - default build, no saas
+#   saas         - proprietary + the saas subproject (build + JUnit only,
+#                  never any runtime/integration testing)
 on:
   workflow_call:
 
@@ -25,7 +31,7 @@ jobs:
       fail-fast: false
       matrix:
         jdk-version: [25]
-        spring-security: [true, false]
+        flavor: [core, proprietary, saas]
     steps:
       - name: Harden Runner
         uses: step-security/harden-runner@ab7a9404c0f3da075243ca237b5fac12c98deaa5 # v2.19.3
@@ -58,7 +64,10 @@ jobs:
       - name: Install Task
         uses: go-task/setup-task@3be4020d41929789a01026e0e427a4321ce0ad44 # v2.0.0
       - name: Check Java formatting (Spotless)
-        if: matrix.jdk-version == 25 && matrix.spring-security == false
+        # Runs once per matrix combination - pick the cheapest leg
+        # (core - no proprietary, no saas) so we don't wait for the
+        # heavier flavors just to fail formatting.
+        if: matrix.jdk-version == 25 && matrix.flavor == 'core'
         id: spotless-check
         run: task backend:format:check
         continue-on-error: true
@@ -67,7 +76,7 @@ jobs:
           MAVEN_PASSWORD: ${{ secrets.MAVEN_PASSWORD }}
           MAVEN_PUBLIC_URL: ${{ secrets.MAVEN_PUBLIC_URL }}
 
-      - name: Comment on Java formatting failure
+      - name: Comment on backend format check failure
         # Only post a comment on PRs. github-script's PR helpers need an
         # issue/PR number, which doesn't exist on merge_group runs.
         if: steps.spotless-check.outcome == 'failure' && github.event_name == 'pull_request'
@@ -78,15 +87,11 @@ jobs:
             const marker = '<!-- java-formatting-check -->';
             const body = [
               marker,
-              '### Java Formatting Check Failed',
+              '### Backend Format Check Failed',
               '',
-              'Your code has formatting issues. Run the following command to fix them:',
+              'There are formatting issues in your Java code that will need to be fixed before they can be merged in.',
               '',
-              '```bash',
-              'task backend:format',
-              '```',
-              '',
-              'Then commit and push the changes.',
+              'Run `task backend:format` to auto-fix, then commit and push the changes.',
             ].join('\n');
             const { data: comments } = await github.rest.issues.listComments({
               owner: context.repo.owner,
@@ -110,33 +115,61 @@ jobs:
               });
             }
 
-      - name: Fail if Java formatting issues found
+      - name: Fail if backend format check failed
         if: steps.spotless-check.outcome == 'failure'
         run: |
           echo "============================================"
-          echo "  Java Formatting Check Failed"
+          echo "  Backend Format Check Failed"
           echo "============================================"
           echo ""
-          echo "Your code has formatting issues."
-          echo "Run the following command to fix them:"
+          echo "There are formatting issues in your Java code"
+          echo "that will need to be fixed before they can be"
+          echo "merged in."
           echo ""
-          echo "  task backend:format"
-          echo ""
-          echo "Then commit and push the changes."
+          echo "Run 'task backend:format' to auto-fix, then"
+          echo "commit and push the changes."
           echo "============================================"
           exit 1
 
-      - name: Build with Gradle and spring security ${{ matrix.spring-security }}
+      - name: Remove backend format check comment on success
+        if: steps.spotless-check.outcome == 'success' && github.event_name == 'pull_request'
+        continue-on-error: true
+        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
+        with:
+          script: |
+            const marker = '<!-- java-formatting-check -->';
+            const { data: comments } = await github.rest.issues.listComments({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: context.issue.number,
+            });
+            const existing = comments.find(c => c.body.includes(marker));
+            if (existing) {
+              await github.rest.issues.deleteComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                comment_id: existing.id,
+              });
+            }
+
+      - name: Build with Gradle (flavor=${{ matrix.flavor }})
+        # STIRLING_FLAVOR is read by settings.gradle and expands into the
+        # right combination of DISABLE_ADDITIONAL_FEATURES + ENABLE_SAAS
+        # so we don't have to set them by hand. The saas flavor pulls in
+        # the app/saas subproject (unit tests only - no runtime tests).
         run: task backend:build:ci
         env:
           MAVEN_USER: ${{ secrets.MAVEN_USER }}
           MAVEN_PASSWORD: ${{ secrets.MAVEN_PASSWORD }}
           MAVEN_PUBLIC_URL: ${{ secrets.MAVEN_PUBLIC_URL }}
-          DISABLE_ADDITIONAL_FEATURES: ${{ matrix.spring-security }}
+          STIRLING_FLAVOR: ${{ matrix.flavor }}
 
       - name: Check Test Reports Exist
         if: always()
         run: |
+          # Common + core + proprietary always build (proprietary is
+          # excluded only at runtime, not from the gradle subproject
+          # graph). Saas builds add a fourth report dir.
           declare -a dirs=(
             "app/core/build/reports/tests/"
             "app/core/build/test-results/"
@@ -145,6 +178,9 @@ jobs:
             "app/proprietary/build/reports/tests/"
             "app/proprietary/build/test-results/"
           )
+          if [ "${{ matrix.flavor }}" = "saas" ]; then
+            dirs+=("app/saas/build/reports/tests/" "app/saas/build/test-results/")
+          fi
           for dir in "${dirs[@]}"; do
             if [ ! -d "$dir" ]; then
               echo "Missing $dir"
@@ -156,7 +192,7 @@ jobs:
         if: always()
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
         with:
-          name: test-reports-jdk-${{ matrix.jdk-version }}-spring-security-${{ matrix.spring-security }}
+          name: test-reports-jdk-${{ matrix.jdk-version }}-flavor-${{ matrix.flavor }}
           path: |
             app/**/build/reports/jacoco/test
             app/**/build/reports/tests/
@@ -166,7 +202,47 @@ jobs:
           retention-days: 3
           if-no-files-found: warn
 
-      - name: Add coverage to PR with spring security ${{ matrix.spring-security }} and JDK ${{ matrix.jdk-version }}
+      - name: Install defusedxml for coverage summary
+        # coverage-summary.py parses JaCoCo XML through defusedxml to
+        # silence security scanners that pattern-match on the stdlib
+        # xml.etree.ElementTree.parse call.
+        if: always() && matrix.flavor == 'saas'
+        run: python -m pip install --quiet defusedxml
+
+      - name: JaCoCo coverage step summary
+        # Only the saas leg posts the JUnit summary - it's a strict
+        # superset of the core + proprietary legs (same .exec files plus
+        # the saas subproject). Posting from all three would mean three
+        # near-identical tables crowding out the aggregate report.
+        if: always() && matrix.flavor == 'saas'
+        run: |
+          python scripts/coverage-summary.py \
+            --title "Backend JUnit coverage (JDK ${{ matrix.jdk-version }})" \
+            --jacoco "common=app/common/build/reports/jacoco/test/jacocoTestReport.xml" \
+            --jacoco "core=app/core/build/reports/jacoco/test/jacocoTestReport.xml" \
+            --jacoco "proprietary=app/proprietary/build/reports/jacoco/test/jacocoTestReport.xml" \
+            --jacoco "saas=app/saas/build/reports/jacoco/test/jacocoTestReport.xml" \
+            --github-step-summary
+
+      - name: Upload raw JUnit .exec for aggregate merge
+        # Same dedup rationale as the summary step: upload from the saas
+        # leg only (the most complete set, includes app/saas/.../test.exec)
+        # so the aggregate workflow merges the union rather than three
+        # overlapping subsets.
+        #
+        # Separate artifact from the HTML reports so the aggregate
+        # workflow can grab just the .exec files with a name pattern
+        # (`jacoco-exec-*`) instead of unpacking the whole test-reports
+        # tarball.
+        if: always() && matrix.flavor == 'saas'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: jacoco-exec-junit-jdk-${{ matrix.jdk-version }}
+          path: app/*/build/jacoco/*.exec
+          retention-days: 7
+          if-no-files-found: warn
+
+      - name: Add coverage to PR (flavor=${{ matrix.flavor }}, JDK=${{ matrix.jdk-version }})
         # The action only supports the pull_request event (it posts a PR comment),
         # so skip it for merge_group runs and workflow_dispatch.
         if: github.event_name == 'pull_request'

.github/workflows/build.yml

@@ -68,6 +68,18 @@ jobs:
     uses: ./.github/workflows/backend-build.yml
     secrets: inherit
 
+  db-migration-test:
+    # Boots the current bootJar against H2 fixtures captured from past
+    # releases (v2.0.0 / v2.5.0 / v2.10.0) and verifies admin login still
+    # works after Hibernate's ddl-auto=update migrates the schema. Gated on
+    # the `project` filter so doc-only PRs skip this ~5-minute job.
+    if: needs.files-changed.outputs.project == 'true'
+    needs: [files-changed]
+    permissions:
+      contents: read
+    uses: ./.github/workflows/db-migration-test.yml
+    secrets: inherit
+
   check-generateOpenApiDocs:
     if: needs.files-changed.outputs.openapi == 'true'
     needs: [files-changed]
@@ -173,6 +185,24 @@ jobs:
     uses: ./.github/workflows/dependency-review.yml
     secrets: inherit
 
+  # Coverage aggregate: merges the JUnit + e2e:live + cucumber .exec
+  # artifacts produced by the jobs above into one report, plus pulls
+  # in vitest + Playwright frontend coverage for the per-area matrix.
+  # `if: always()` so a producer failing partway still gets credit
+  # for whatever did record. Advisory only - intentionally NOT in
+  # all-checks-passed, so a flaky aggregate run never blocks merging.
+  coverage-aggregate:
+    if: always()
+    needs:
+      - build
+      - playwright-e2e-live
+      - docker-compose-tests
+      - frontend-validation
+    permissions:
+      contents: read
+    uses: ./.github/workflows/coverage-aggregate.yml
+    secrets: inherit
+
   # Single status check that branch protection should mark as required.
   # Succeeds when every upstream job is either `success` or `skipped` (path-
   # gated jobs that didn't apply this run). Any `failure` or `cancelled`
@@ -184,6 +214,7 @@ jobs:
     needs:
       - files-changed
       - build
+      - db-migration-test
       - check-generateOpenApiDocs
       - frontend-validation
       - playwright-e2e
@@ -208,6 +239,7 @@ jobs:
           RESULTS: |
             files-changed=${{ needs.files-changed.result }}
             build=${{ needs.build.result }}
+            db-migration-test=${{ needs.db-migration-test.result }}
             check-generateOpenApiDocs=${{ needs.check-generateOpenApiDocs.result }}
             frontend-validation=${{ needs.frontend-validation.result }}
             playwright-e2e=${{ needs.playwright-e2e.result }}

.github/workflows/coverage-aggregate.yml

@@ -0,0 +1,230 @@
+name: Aggregate backend coverage
+
+# Reusable workflow called from build.yml after every backend coverage
+# producer (JUnit, e2e:live, cucumber) has run. Downloads each job's raw
+# .exec, merges them into one JaCoCo report, and posts a combined step
+# summary alongside the per-source ones.
+#
+# Kept separate from the per-source jobs so:
+#   - the per-source jobs stay fast and independent (no cross-job waits)
+#   - this job can `if: always()` and still produce something useful when
+#     one of the producers fails partway through
+#   - frontend producers can be added later without touching the
+#     producers themselves
+on:
+  workflow_call:
+
+permissions:
+  contents: read
+
+jobs:
+  pick:
+    uses: ./.github/workflows/_runner-pick.yml
+
+  aggregate:
+    needs: pick
+    runs-on: ${{ needs.pick.outputs.is_fork == 'true' && 'ubuntu-latest' || 'depot-ubuntu-24.04-4' }}
+    timeout-minutes: 15
+    steps:
+      - name: Harden Runner
+        uses: step-security/harden-runner@ab7a9404c0f3da075243ca237b5fac12c98deaa5 # v2.19.3
+        with:
+          egress-policy: audit
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+      - name: Set up JDK 25
+        uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
+        with:
+          java-version: "25"
+          distribution: "temurin"
+
+      - name: Cache Gradle dependency artifacts
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
+        with:
+          path: |
+            ~/.gradle/wrapper
+            ~/.gradle/caches/modules-2/files-2.1
+            ~/.gradle/caches/modules-2/metadata-2.*
+          key: gradle-deps-${{ 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 Python
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.12"
+
+      - name: Install defusedxml for coverage scripts
+        # Both coverage-summary.py and coverage-matrix.py parse JaCoCo
+        # XML through defusedxml - see the script headers for context.
+        run: python -m pip install --quiet defusedxml
+
+      # Pattern matches every artifact this PR's producers might upload:
+      #   jacoco-exec-junit-jdk-25            (uploaded only by the saas
+      #                                         leg of backend-build, which
+      #                                         is a strict superset of the
+      #                                         core + proprietary legs)
+      #   jacoco-exec-e2e-live
+      #   jacoco-exec-cucumber
+      # Each lands as a sibling dir under coverage-execs/, with the .exec
+      # files preserving their original relative paths.
+      - name: Download all .exec artifacts
+        uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v6.0.0
+        with:
+          pattern: jacoco-exec-*
+          path: coverage-execs/
+          merge-multiple: false
+        continue-on-error: true
+
+      - name: Inventory .exec files
+        id: inventory
+        # Splits the downloaded artifacts into two buckets:
+        #   * e2e-only  = cucumber + Playwright live (user-flow coverage)
+        #   * all       = the above plus JUnit (everything we test)
+        #
+        # Bucketing is by artifact-name prefix: download-artifact preserves
+        # the artifact name as the top-level dir, so JUnit's `.exec`s live
+        # under coverage-execs/jacoco-exec-junit-*/... while the others
+        # are under coverage-execs/jacoco-exec-{e2e-live,cucumber}/...
+        #
+        # If nothing was uploaded (e.g. all producers crashed before
+        # writing) we exit gracefully so this advisory job never fails CI.
+        run: |
+          mapfile -t all_execs < <(find coverage-execs -name '*.exec' -type f | sort)
+          mapfile -t e2e_execs < <(find coverage-execs -name '*.exec' -type f -not -path '*/jacoco-exec-junit-*' | sort)
+          if [ "${#all_execs[@]}" -eq 0 ]; then
+            echo "::warning::No .exec artifacts found - skipping aggregate report"
+            echo "found_all=false" >> "$GITHUB_OUTPUT"
+            echo "found_e2e=false" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+          printf 'All %d .exec files:\n' "${#all_execs[@]}"
+          printf '  %s\n' "${all_execs[@]}"
+          IFS=','; all_joined="${all_execs[*]}"
+          echo "files_all=$all_joined" >> "$GITHUB_OUTPUT"
+          echo "found_all=true" >> "$GITHUB_OUTPUT"
+          if [ "${#e2e_execs[@]}" -eq 0 ]; then
+            echo "::notice::No e2e/cucumber .exec files - e2e-only report will be skipped"
+            echo "found_e2e=false" >> "$GITHUB_OUTPUT"
+          else
+            printf 'E2E-only %d .exec files:\n' "${#e2e_execs[@]}"
+            printf '  %s\n' "${e2e_execs[@]}"
+            unset IFS
+            IFS=','; e2e_joined="${e2e_execs[*]}"
+            echo "files_e2e=$e2e_joined" >> "$GITHUB_OUTPUT"
+            echo "found_e2e=true" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Compile classes for JaCoCo class lookup
+        # jacocoReportFromExec only needs the compiled .class files
+        # under each subproject's build/classes/java/main/. `classes`
+        # (compileJava + processResources) is enough; we skipped the
+        # heavier `assemble` to avoid building bootJar / fat jars that
+        # add 60+ seconds per run for no gain to the report.
+        if: steps.inventory.outputs.found_all == 'true'
+        run: ./gradlew classes -PnoSpotless
+
+      - name: Generate e2e-only JaCoCo report
+        # "Real user-flow" coverage: only counts code reached by an actual
+        # HTTP request from cucumber or live Playwright. Useful for
+        # questions like "how much of our backend does a user actually
+        # hit?". Skipped when neither producer uploaded a .exec.
+        if: steps.inventory.outputs.found_e2e == 'true'
+        run: |
+          ./gradlew jacocoReportFromExec \
+            -PexecFile="${{ steps.inventory.outputs.files_e2e }}" \
+            -PreportDir=build/reports/jacoco/aggregate-e2e \
+            -PnoSpotless
+
+      - name: Generate combined JaCoCo report (everything)
+        if: steps.inventory.outputs.found_all == 'true'
+        run: |
+          ./gradlew jacocoReportFromExec \
+            -PexecFile="${{ steps.inventory.outputs.files_all }}" \
+            -PreportDir=build/reports/jacoco/aggregate-all \
+            -PnoSpotless
+
+      - name: E2E-only step summary
+        # Rendered first so it gets prime real estate in the Summary
+        # tab - this is the number most readers actually want
+        # ("how much of the backend do real user flows cover?").
+        if: steps.inventory.outputs.found_e2e == 'true'
+        run: |
+          python scripts/coverage-summary.py \
+            --title "Real user-flow backend coverage (e2e:live + cucumber)" \
+            --jacoco "merged=build/reports/jacoco/aggregate-e2e/jacocoTestReport.xml" \
+            --github-step-summary
+
+      - name: ALL-sources step summary
+        # Separate call (not a multi-input one) because the helper's
+        # rightmost "Aggregate" column would sum the two reports - which
+        # is meaningless when one is a strict superset of the other.
+        if: steps.inventory.outputs.found_all == 'true'
+        run: |
+          python scripts/coverage-summary.py \
+            --title "Combined backend coverage (JUnit + e2e:live + cucumber)" \
+            --jacoco "merged=build/reports/jacoco/aggregate-all/jacocoTestReport.xml" \
+            --github-step-summary
+
+      - name: Upload combined aggregate report
+        if: steps.inventory.outputs.found_all == 'true'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: jacoco-aggregate-all-${{ github.run_id }}
+          path: build/reports/jacoco/aggregate-all/
+          retention-days: 14
+
+      - name: Upload e2e-only aggregate report
+        if: steps.inventory.outputs.found_e2e == 'true'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: jacoco-aggregate-e2e-${{ github.run_id }}
+          path: build/reports/jacoco/aggregate-e2e/
+          retention-days: 14
+
+      # --------------------------------------------------------------
+      # Per-area matrix: rolls backend + frontend coverage into one
+      # table indexed by core/proprietary/saas/desktop. Pulls the
+      # frontend artifacts now (after the JaCoCo step has done its
+      # work) so the per-source backend summaries still render first
+      # even if the matrix step fails.
+      # --------------------------------------------------------------
+      - name: Download vitest coverage artifact
+        # frontend-validation uploads as `frontend-coverage`. Tolerate
+        # absence so a backend-only PR still produces the matrix with
+        # just backend rows populated.
+        if: always()
+        uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v6.0.0
+        with:
+          name: frontend-coverage
+          path: matrix-inputs/vitest/
+        continue-on-error: true
+
+      - name: Download Playwright frontend coverage artifact
+        # e2e-live uploads as `playwright-frontend-coverage-<run_id>`.
+        # Same tolerance as vitest - matrix script handles missing inputs.
+        if: always()
+        uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v6.0.0
+        with:
+          name: playwright-frontend-coverage-${{ github.run_id }}
+          path: matrix-inputs/playwright/
+        continue-on-error: true
+
+      - name: Coverage matrix step summary
+        if: always()
+        # Matrix references the two aggregate JaCoCo XMLs (already
+        # generated above) plus whichever frontend artifacts landed.
+        # Every input is optional; missing ones render as "-".
+        run: |
+          python scripts/coverage-matrix.py \
+            ${{ steps.inventory.outputs.found_all == 'true' && '--jacoco-all build/reports/jacoco/aggregate-all/jacocoTestReport.xml' || '' }} \
+            ${{ steps.inventory.outputs.found_e2e == 'true' && '--jacoco-e2e build/reports/jacoco/aggregate-e2e/jacocoTestReport.xml' || '' }} \
+            --vitest matrix-inputs/vitest/coverage-summary.json \
+            --playwright-frontend matrix-inputs/playwright/coverage-pw-summary/coverage-summary.json \
+            --title "Coverage matrix (per-area, e2e vs all)" \
+            --github-step-summary

.github/workflows/db-migration-test.yml

@@ -0,0 +1,93 @@
+name: DB migration smoke test
+
+# Boots the current Stirling-PDF JAR against H2 fixtures captured from past
+# releases (v2.0.0 / v2.5.0 / v2.10.0) and verifies admin login still works.
+# Catches schema changes that would break existing user databases under
+# Hibernate's `ddl-auto=update` upgrade path.
+
+on:
+  workflow_call:
+
+permissions:
+  contents: read
+
+jobs:
+  pick:
+    uses: ./.github/workflows/_runner-pick.yml
+
+  migration-test:
+    needs: pick
+    runs-on: ${{ needs.pick.outputs.is_fork == 'true' && 'ubuntu-latest' || 'depot-ubuntu-24.04-8' }}
+    timeout-minutes: 30
+    env:
+      DEPOT_TOKEN: ${{ secrets.DEPOT_TOKEN }}
+    steps:
+      - name: Harden Runner
+        uses: step-security/harden-runner@ab7a9404c0f3da075243ca237b5fac12c98deaa5 # v2.19.3
+        with:
+          egress-policy: audit
+
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+      - name: Set up JDK 25
+        uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
+        with:
+          java-version: 25
+          distribution: temurin
+
+      - name: Cache Gradle dependency artifacts
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
+        with:
+          path: |
+            ~/.gradle/wrapper
+            ~/.gradle/caches/modules-2/files-2.1
+            ~/.gradle/caches/modules-2/metadata-2.*
+          key: gradle-deps-${{ 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
+
+      # No `-PnoSpotless` here yet because the upstream cache layer matches the
+      # backend build's; reuse keeps cold-cache cost identical.
+      - name: Build Stirling-PDF JAR
+        env:
+          MAVEN_USER: ${{ secrets.MAVEN_USER }}
+          MAVEN_PASSWORD: ${{ secrets.MAVEN_PASSWORD }}
+          MAVEN_PUBLIC_URL: ${{ secrets.MAVEN_PUBLIC_URL }}
+        run: ./gradlew :stirling-pdf:bootJar -PnoSpotless --no-daemon
+
+      - name: Locate built JAR
+        id: jar
+        run: |
+          jar=$(find app/core/build/libs -maxdepth 1 -name 'Stirling-PDF*.jar' -o -name 'stirling-pdf*.jar' 2>/dev/null \
+              | grep -vE '(-plain|-sources)\.jar$' | head -n 1)
+          if [[ -z "$jar" ]]; then
+            echo "::error::No JAR under app/core/build/libs"
+            ls -lah app/core/build/libs || true
+            exit 1
+          fi
+          # Absolute path - the migration script pushd's into a temp workdir
+          # before invoking java, which would dangle a relative path.
+          jar=$(realpath "$jar")
+          echo "path=$jar" >> "$GITHUB_OUTPUT"
+          echo "Built JAR: $jar"
+
+      - name: Run migration smoke test
+        env:
+          STIRLING_JAR: ${{ steps.jar.outputs.path }}
+        run: bash scripts/db-migration/run-migration-test.sh
+
+      - name: Upload app logs on failure
+        if: failure()
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: db-migration-app-logs
+          # Path matches the preserved workdir in run-migration-test.sh -
+          # only failing fixtures leave a directory behind.
+          path: /tmp/stirling-migration-failed-*/app.log
+          retention-days: 7
+          if-no-files-found: warn

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

@@ -87,6 +87,12 @@ jobs:
         run: |
           pip install --require-hashes --only-binary=:all: -r ./testing/cucumber/requirements.txt
 
+      - name: Extract JaCoCo agent for cucumber coverage
+        # Stages build/jacoco/jacocoagent.jar where the coverage override
+        # file bind-mounts it into the cucumber container. The agent jar
+        # never goes into the published image - this is host-only.
+        run: ./gradlew copyJacocoAgent -PnoSpotless
+
       - name: Run Docker Compose Tests
         run: |
           chmod +x ./testing/test_webpages.sh
@@ -98,6 +104,62 @@ jobs:
           MAVEN_PASSWORD: ${{ secrets.MAVEN_PASSWORD }}
           MAVEN_PUBLIC_URL: ${{ secrets.MAVEN_PUBLIC_URL }}
           DOCKER_BASE_CHANGED: ${{ inputs.docker-base-changed }}
+          # Tells test.sh to layer testing/compose/docker-compose-coverage.override.yml
+          # over the cucumber compose so the container starts with the
+          # JaCoCo agent attached via JAVA_CUSTOM_OPTS.
+          STIRLING_PDF_TEST_COVERAGE: "1"
+
+      - name: Generate cucumber JaCoCo report
+        # `if: always()` so a behave failure still produces partial
+        # coverage from whatever endpoints did run. The exec file only
+        # exists when the container shut down cleanly - guard so the step
+        # is silent on the (rare) crash path.
+        if: always()
+        id: cucumber-coverage
+        run: |
+          if [ -s testing/cucumber-coverage/cucumber.exec ]; then
+            ./gradlew jacocoReportFromExec \
+              -PexecFile=testing/cucumber-coverage/cucumber.exec \
+              -PreportDir=build/reports/jacoco/cucumber \
+              -PnoSpotless
+            echo "report=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "::warning::No cucumber .exec at testing/cucumber-coverage/cucumber.exec (container may have crashed before flushing)"
+            echo "report=false" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Install defusedxml for coverage summary
+        # coverage-summary.py parses JaCoCo XML through defusedxml -
+        # see the script header for context.
+        if: always() && steps.cucumber-coverage.outputs.report == 'true'
+        run: python -m pip install --quiet defusedxml
+
+      - name: Cucumber coverage step summary
+        if: always() && steps.cucumber-coverage.outputs.report == 'true'
+        run: |
+          python scripts/coverage-summary.py \
+            --title "Cucumber (docker) JaCoCo coverage" \
+            --jacoco "cucumber=build/reports/jacoco/cucumber/jacocoTestReport.xml" \
+            --github-step-summary
+
+      - name: Upload cucumber JaCoCo report
+        if: always() && steps.cucumber-coverage.outputs.report == 'true'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: jacoco-cucumber-${{ github.run_id }}
+          path: build/reports/jacoco/cucumber/
+          retention-days: 7
+
+      - name: Upload raw cucumber .exec for aggregate merge
+        # Picked up by the coverage-aggregate workflow via the
+        # `jacoco-exec-*` artifact name pattern.
+        if: always() && steps.cucumber-coverage.outputs.report == 'true'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: jacoco-exec-cucumber
+          path: testing/cucumber-coverage/cucumber.exec
+          retention-days: 7
+          if-no-files-found: warn
 
       - name: Upload Cucumber Report
         if: always()

.github/workflows/e2e-live.yml

@@ -49,9 +49,132 @@ jobs:
         env:
           VITE_BUILD_FOR_PREVIEW: "1"
         run: task frontend:build
-      - name: Run live E2E tests (chromium)
+      - name: Run live E2E tests (chromium) with coverage
         id: live-tests
+        env:
+          # Attaches the JaCoCo agent to the bootRun JVM (see
+          # .taskfiles/e2e.yml live:backend). The .exec gets flushed on
+          # graceful shutdown when the runner traps EXIT/INT/TERM, so the
+          # report step below sees a populated file.
+          COVERAGE: "1"
+          # Tells the Playwright fixture (test-base.ts) to capture per-test
+          # V8 JS coverage. Raw dumps land under
+          # .test-state/playwright/coverage-pw/ for the post-process step
+          # to aggregate. Chromium-only - other engines silently skip.
+          PW_COVERAGE: "1"
         run: task e2e:live
+      - name: Generate JaCoCo report from e2e:live .exec
+        if: always()
+        id: live-coverage
+        # `if: always()` so even a failed test run still produces a
+        # report from whatever flows did exercise the backend before
+        # the failure. The task itself tolerates a missing .exec
+        # (jacoco emits an empty report rather than crashing) but we
+        # guard with `test -s` to keep the job log clean.
+        run: |
+          if [ -s .test-state/playwright/jacoco.exec ]; then
+            ./gradlew jacocoReportFromExec \
+              -PexecFile=.test-state/playwright/jacoco.exec \
+              -PreportDir=build/reports/jacoco/e2e-live \
+              -PnoSpotless
+            echo "report=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "::warning::No e2e:live .exec found at .test-state/playwright/jacoco.exec; skipping report"
+            echo "report=false" >> "$GITHUB_OUTPUT"
+          fi
+      - name: Set up Python for coverage summary
+        if: always() && steps.live-coverage.outputs.report == 'true'
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.12"
+      - name: Install defusedxml for coverage summary
+        # coverage-summary.py uses defusedxml instead of stdlib xml.etree
+        # to dodge XXE / billion-laughs scanner findings.
+        if: always() && steps.live-coverage.outputs.report == 'true'
+        run: python -m pip install --quiet defusedxml
+      - name: e2e:live coverage step summary
+        if: always() && steps.live-coverage.outputs.report == 'true'
+        run: |
+          python scripts/coverage-summary.py \
+            --title "Playwright (live backend) JaCoCo coverage" \
+            --jacoco "e2e-live=build/reports/jacoco/e2e-live/jacocoTestReport.xml" \
+            --github-step-summary
+      - name: Upload e2e:live JaCoCo report
+        if: always() && steps.live-coverage.outputs.report == 'true'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: jacoco-e2e-live-${{ github.run_id }}
+          path: build/reports/jacoco/e2e-live/
+          retention-days: 7
+
+      - name: Upload raw e2e:live .exec for aggregate merge
+        # Picked up by the coverage-aggregate workflow via the
+        # `jacoco-exec-*` artifact name pattern.
+        if: always() && steps.live-coverage.outputs.report == 'true'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: jacoco-exec-e2e-live
+          path: .test-state/playwright/jacoco.exec
+          retention-days: 7
+          if-no-files-found: warn
+
+      - name: Set up Python for frontend coverage summary
+        # Separate from the backend-coverage python step because the
+        # frontend path doesn't depend on a JaCoCo report - it produces
+        # a summary even on backend failure, as long as some Playwright
+        # tests ran far enough to dump V8 coverage.
+        if: always()
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.12"
+
+      - name: Install defusedxml for frontend coverage summary
+        # Idempotent re-install: the backend-coverage step may have
+        # installed it already, but this leg can run on its own when the
+        # backend report step skips (e.g. .exec missing).
+        if: always()
+        run: python -m pip install --quiet defusedxml
+
+      - name: Aggregate Playwright frontend (V8) coverage
+        # Rolls per-test V8 dumps from the test-base fixture into one
+        # vitest-shaped coverage-summary.json. Tolerates a missing dump
+        # dir (firefox/webkit runs, or a failure before any test got
+        # far enough to dump).
+        if: always()
+        id: pw-frontend-coverage
+        run: |
+          if [ -d .test-state/playwright/coverage-pw ] && \
+             find .test-state/playwright/coverage-pw -name '*.json' -type f | grep -q .; then
+            python scripts/playwright-coverage-summary.py \
+              .test-state/playwright/coverage-pw \
+              --out .test-state/playwright/coverage-pw-summary/coverage-summary.json
+            echo "summary=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "::notice::No Playwright frontend coverage dumps found (chromium-only feature)"
+            echo "summary=false" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Playwright frontend coverage step summary
+        if: always() && steps.pw-frontend-coverage.outputs.summary == 'true'
+        run: |
+          python scripts/coverage-summary.py \
+            --title "Playwright (live) frontend coverage" \
+            --vitest .test-state/playwright/coverage-pw-summary/coverage-summary.json \
+            --github-step-summary
+
+      - name: Upload Playwright frontend coverage
+        # Bundle both the aggregated summary and the raw V8 dumps so
+        # someone debugging "why is this function showing as covered"
+        # can trace it back to the source dump.
+        if: always() && steps.pw-frontend-coverage.outputs.summary == 'true'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: playwright-frontend-coverage-${{ github.run_id }}
+          path: |
+            .test-state/playwright/coverage-pw-summary/
+            .test-state/playwright/coverage-pw/
+          retention-days: 7
+
       - name: Print backend log on failure
         if: failure() && steps.live-tests.conclusion == 'failure'
         run: |

.github/workflows/frontend-backend-licenses-update.yml

@@ -110,8 +110,8 @@ jobs:
           NPM_CONFIG_IGNORE_SCRIPTS: "true"
         working-directory: frontend
         run: |
-          mkdir -p src/assets
-          npx --yes license-report --only=prod --output=json > src/assets/3rdPartyLicenses.json
+          mkdir -p editor/src/assets
+          npx --yes license-report --only=prod --output=json > editor/src/assets/3rdPartyLicenses.json
 
       - name: Postprocess with project script (BASE version)
         if: github.event_name == 'pull_request' && github.event.pull_request.head.repo.fork == true

.github/workflows/frontend-validation.yml

@@ -109,6 +109,41 @@ jobs:
                 comment_id: existing.id,
               });
             }
+      - name: Vitest coverage
+        # Separate from `frontend:check:all` so the quality-gate run stays
+        # uninstrumented (faster signal) and coverage stays an informational
+        # follow-up. Continue-on-error keeps the workflow green even when
+        # a handful of test files refuse to import (e.g. missing icon
+        # specifiers) - the summary still gets posted with whatever
+        # vitest managed to instrument.
+        id: frontend-coverage
+        continue-on-error: true
+        run: task frontend:test:coverage
+      - name: Set up Python for coverage summary
+        if: always()
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.12"
+      - name: Install defusedxml for coverage summary
+        # See coverage-summary.py header - it parses XML through defusedxml
+        # to dodge the stdlib parser's exposure to XXE / billion-laughs.
+        if: always()
+        run: python -m pip install --quiet defusedxml
+      - name: Vitest coverage step summary
+        if: always()
+        run: |
+          python scripts/coverage-summary.py \
+            --title "Frontend Vitest coverage" \
+            --vitest frontend/editor/coverage/coverage-summary.json \
+            --github-step-summary
+      - name: Upload vitest coverage report
+        if: always()
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: frontend-coverage
+          path: frontend/editor/coverage/
+          retention-days: 7
+          if-no-files-found: warn
       - name: Upload frontend build artifacts
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
         with:

.github/workflows/multiOSReleases.yml

@@ -442,10 +442,6 @@ jobs:
           echo "Generated tauri.windows.conf.json (alias masked):"
           sed "s/${KEYPAIR_ALIAS}/***/g" ./frontend/editor/src-tauri/tauri.windows.conf.json
 
-      - name: Sign JPDFium dylibs inside bootJar (macOS only)
-        if: matrix.platform == 'macos-15' && env.APPLE_CERTIFICATE != ''
-        run: bash frontend/scripts/sign-jpdfium-dylibs-in-bootjar.sh
-
       - name: Import release GPG signing key (Linux)
         if: matrix.platform == 'ubuntu-22.04' && env.RELEASE_GPG_PRIVATE_KEY != '' && (github.event_name == 'release' || (github.event_name == 'workflow_dispatch' && github.event.inputs.sign != 'false') || github.ref == 'refs/heads/V2-master')
         run: |
@@ -495,6 +491,7 @@ jobs:
           projectPath: ./frontend/editor
           tauriScript: npx tauri
           args: ${{ matrix.args }}
+          updaterJsonKeepUniversal: true
 
       - name: Clear release GPG key from runner keyring (Linux)
         if: always() && matrix.platform == 'ubuntu-22.04' && env.RELEASE_GPG_PRIVATE_KEY != '' && (github.event_name == 'release' || (github.event_name == 'workflow_dispatch' && github.event.inputs.sign != 'false') || github.ref == 'refs/heads/V2-master')
@@ -586,21 +583,37 @@ jobs:
         if: always() && steps.digicert-setup.conclusion != 'failure'
         shell: bash
         run: |
-          mkdir -p ./dist
+          # Absolute dist path so the cd below can't break the copy targets.
+          DIST="$GITHUB_WORKSPACE/dist"
+          mkdir -p "$DIST"
           cd ./frontend/editor/src-tauri/target
 
-          # Find and rename artifacts based on platform
+          echo "=== tauri bundle artifacts ==="
+          find . -path "*/bundle/*" \( -name "*.msi" -o -name "*.deb" \
+            -o -name "*.rpm" -o -name "*.AppImage" -o -name "*.dmg" \
+            -o -name "*.app.tar.gz" -o -name "*.sig" \) 2>/dev/null | sort || true
+          echo "=============================="
+
+          # createUpdaterArtifacts:true signs the native installers in place;
+          # each <bundle> ships with a sibling <bundle>.sig consumed by latest.json.
           if [ "${{ matrix.platform }}" = "windows-latest" ]; then
-            # Only ship the MSI installer on Windows. The loose exe and WiX toolset exes
-            # are not the user-facing installer - the MSI contains the signed inner exe.
-            find . -name "*.msi" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.msi" \;
+            find . -name "*.msi" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.msi" \;
+            find . -name "*.msi.sig" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.msi.sig" \;
           elif [ "${{ matrix.platform }}" = "macos-15" ]; then
-            find . -name "*.dmg" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.dmg" \;
-            find . -name "*.app" -exec cp -r {} "../../../dist/Stirling-PDF-${{ matrix.name }}.app" \;
+            # DMG = manual install; .app.tar.gz (+ .sig) = updater payload.
+            # Raw .app is intentionally not shipped (hundreds of MB of uncompressed input).
+            find . -name "*.dmg" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.dmg" \;
+            find . -name "*.app.tar.gz" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.app.tar.gz" \;
+            find . -name "*.app.tar.gz.sig" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.app.tar.gz.sig" \;
           else
-            find . -name "*.deb" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.deb" \;
-            find . -name "*.rpm" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.rpm" \;
-            find . -name "*.AppImage" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.AppImage" \;
+            # The raw .AppImage IS its updater payload (signed -> .AppImage.sig),
+            # not a .tar.gz wrapper - that's only produced under v1Compatible.
+            find . -name "*.deb" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.deb" \;
+            find . -name "*.deb.sig" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.deb.sig" \;
+            find . -name "*.rpm" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.rpm" \;
+            find . -name "*.rpm.sig" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.rpm.sig" \;
+            find . -name "*.AppImage" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.AppImage" \;
+            find . -name "*.AppImage.sig" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.AppImage.sig" \;
           fi
 
       - name: Upload build artifacts
@@ -611,8 +624,7 @@ jobs:
           path: ./dist/*
           retention-days: 1
 
-  create-release:
-    if: (github.event_name == 'workflow_dispatch' && github.event.inputs.test_mode != 'true') || github.event_name == 'release' || github.ref == 'refs/heads/V2-master'
+  collect-and-release:
     needs: [pick, determine-matrix, build, build-jars]
     runs-on: ${{ needs.pick.outputs.is_fork == 'true' && 'ubuntu-latest' || 'depot-ubuntu-24.04-4' }}
     permissions:
@@ -623,6 +635,16 @@ jobs:
         with:
           egress-policy: audit
 
+      # Sparse-check out the verifier + pubkey before the artifact downloads
+      # so the checkout cannot clobber ./artifacts.
+      - name: Checkout updater verifier
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          sparse-checkout: |
+            .github/scripts/verify-updater-signatures.py
+            frontend/editor/src-tauri/tauri.conf.json
+          sparse-checkout-cone-mode: false
+
       - name: Download all Tauri artifacts
         uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
         with:
@@ -650,17 +672,146 @@ jobs:
       - name: Display structure of downloaded files
         run: ls -R ./artifacts
 
+      # tauri-action only emits latest.json when it also publishes the release
+      # (tagName/releaseId set). We publish separately via action-gh-release,
+      # so build latest.json here from the per-platform .sig files.
+      - name: Generate updater latest.json
+        env:
+          VERSION: ${{ needs.determine-matrix.outputs.version }}
+          TAG: v${{ needs.determine-matrix.outputs.version }}
+          REPO: ${{ github.repository }}
+        run: |
+          python3 - << 'PYEOF'
+          import json, os, sys
+          from pathlib import Path
+          from datetime import datetime, timezone
+
+          VERSION = os.environ['VERSION']
+          TAG = os.environ['TAG']
+          REPO = os.environ['REPO']
+
+          ART = Path('./artifacts/tauri')
+
+          # Tauri updater looks up {os}-{arch}-{installer} (e.g. linux-x86_64-deb)
+          # before bare {os}-{arch}, so per-format Linux keys let deb/rpm/appimage
+          # each self-update from their matching file. macOS universal serves both
+          # arches from the one .app.tar.gz.
+          PLATFORM_MAP = [
+              {
+                  'bundles': ['Stirling-PDF-linux-x86_64.deb'],
+                  'targets': ['linux-x86_64-deb'],
+              },
+              {
+                  'bundles': ['Stirling-PDF-linux-x86_64.rpm'],
+                  'targets': ['linux-x86_64-rpm'],
+              },
+              {
+                  'bundles': ['Stirling-PDF-linux-x86_64.AppImage'],
+                  'targets': ['linux-x86_64-appimage'],
+              },
+              {
+                  'bundles': ['Stirling-PDF-windows-x86_64.msi'],
+                  'targets': ['windows-x86_64-msi', 'windows-x86_64'],
+              },
+              {
+                  'bundles': ['Stirling-PDF-macos-universal.app.tar.gz'],
+                  'targets': ['darwin-x86_64', 'darwin-aarch64'],
+              },
+          ]
+
+          # rglob() because download-artifact varies layout: one artifact -> flat,
+          # many -> nested under <artifact-name>/.
+          def find_signed(name):
+              for bundle_path in sorted(ART.rglob(name)):
+                  sig_path = bundle_path.with_name(bundle_path.name + '.sig')
+                  if sig_path.exists():
+                      return bundle_path, sig_path
+              return None
+
+          platforms = {}
+          skipped = []
+          for entry in PLATFORM_MAP:
+              picked = None
+              for name in entry['bundles']:
+                  picked = find_signed(name)
+                  if picked:
+                      break
+              if not picked:
+                  skipped.append(
+                      f"{entry['targets']} (no signed bundle among "
+                      f"{entry['bundles']} - TAURI_SIGNING_PRIVATE_KEY unset "
+                      f"or createUpdaterArtifacts disabled?)"
+                  )
+                  continue
+              bundle_path, sig_path = picked
+              signature = sig_path.read_text(encoding='utf-8').strip()
+              url = f"https://github.com/{REPO}/releases/download/{TAG}/{bundle_path.name}"
+              for target in entry['targets']:
+                  platforms[target] = {'signature': signature, 'url': url}
+              print(f"Added {entry['targets']} from {bundle_path.name}")
+
+          if skipped:
+              print("Skipped platforms:")
+              for s in skipped:
+                  print(f"  - {s}")
+
+          if not platforms:
+              print(
+                  "WARN: no signed updater bundles found - "
+                  "skipping latest.json generation"
+              )
+              sys.exit(0)
+
+          manifest = {
+              'version': VERSION,
+              'notes': f"See https://github.com/{REPO}/releases/tag/{TAG}",
+              'pub_date': datetime.now(timezone.utc).strftime('%Y-%m-%dT%H:%M:%SZ'),
+              'platforms': platforms,
+          }
+
+          out = Path('./artifacts/latest.json')
+          out.write_text(json.dumps(manifest, indent=2) + '\n', encoding='utf-8')
+          print(f"Generated {out} with platforms: {sorted(platforms.keys())}")
+          PYEOF
+
+      - name: Upload merged artifacts for review
+        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
+        with:
+          name: release-artifacts
+          path: ./artifacts/
+          retention-days: 7
+
+      # Gate publish on valid updater sigs. Runs after the review upload (so
+      # artifacts survive for debugging) and before action-gh-release.
+      - name: Verify updater signatures
+        run: |
+          python3 -m pip install --quiet 'cryptography==44.0.0'
+          python3 .github/scripts/verify-updater-signatures.py \
+            ./artifacts/tauri frontend/editor/src-tauri/tauri.conf.json
+
+      # workflow_dispatch path requires platform=='all' so a single-platform
+      # dispatch can't overwrite an existing release's full latest.json with a
+      # partial one (action-gh-release defaults overwrite_files:true).
+      # release / V2-master always build the full matrix so no extra guard needed.
+      # fail_on_unmatched_files makes a missing latest.json or installer fail loudly
+      # instead of silently shipping a broken auto-update.
       - name: Upload binaries to Release
+        if: (github.event_name == 'workflow_dispatch' && github.event.inputs.test_mode != 'true' && github.event.inputs.platform == 'all') || github.event_name == 'release' || github.ref == 'refs/heads/V2-master'
         uses: softprops/action-gh-release@b4309332981a82ec1c5618f44dd2e27cc8bfbfda # v3.0.0
         with:
           tag_name: v${{ needs.determine-matrix.outputs.version }}
           generate_release_notes: true
+          fail_on_unmatched_files: true
+          # Installers + updater payloads + manifest. .sig contents are embedded
+          # in latest.json so the .sig files themselves are not uploaded.
           files: |
             ./artifacts/**/*.jar
             ./artifacts/**/*.msi
             ./artifacts/**/*.dmg
+            ./artifacts/**/*.app.tar.gz
             ./artifacts/**/*.deb
             ./artifacts/**/*.rpm
             ./artifacts/**/*.AppImage
+            ./artifacts/latest.json
           draft: false
           prerelease: false

.github/workflows/tauri-build.yml

@@ -157,6 +157,9 @@ jobs:
           JPDFIUM_PLATFORMS: ${{ matrix.jpdfium_platforms }}
         run: task desktop:prepare
 
+      - name: Run Tauri/Cargo tests
+        run: task desktop:test
+
       # DigiCert KeyLocker Setup (Cloud HSM)
       - name: Setup DigiCert KeyLocker
         id: digicert-setup
@@ -268,10 +271,6 @@ jobs:
           echo "APPLE_SIGNING_IDENTITY=$CERT_ID" >> $GITHUB_ENV
           echo "Certificate imported successfully."
 
-      - name: Sign JPDFium dylibs inside bootJar (macOS only)
-        if: matrix.platform == 'macos-15' && env.APPLE_CERTIFICATE != ''
-        run: bash frontend/scripts/sign-jpdfium-dylibs-in-bootjar.sh
-
       - name: Check DMG creation dependencies (macOS only)
         if: matrix.platform == 'macos-15'
         run: |
@@ -417,20 +416,22 @@ jobs:
       - name: Rename artifacts
         shell: bash
         run: |
-          mkdir -p ./dist
+          # Absolute dist path so the cd below can't break the copy targets.
+          DIST="$GITHUB_WORKSPACE/dist"
+          mkdir -p "$DIST"
           cd ./frontend/editor/src-tauri/target
 
           # Find and rename artifacts based on platform
           if [ "${{ matrix.platform }}" = "windows-latest" ]; then
             # Only ship the MSI installer. The loose exe and WiX toolset exes
             # are not the user-facing installer - the MSI contains the signed inner exe.
-            find . -name "*.msi" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.msi" \;
+            find . -name "*.msi" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.msi" \;
           elif [ "${{ matrix.platform }}" = "macos-15" ]; then
-            find . -name "*.dmg" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.dmg" \;
+            find . -name "*.dmg" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.dmg" \;
           else
-            find . -name "*.deb" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.deb" \;
-            find . -name "*.rpm" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.rpm" \;
-            find . -name "*.AppImage" -exec cp {} "../../../dist/Stirling-PDF-${{ matrix.name }}.AppImage" \;
+            find . -name "*.deb" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.deb" \;
+            find . -name "*.rpm" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.rpm" \;
+            find . -name "*.AppImage" -exec cp {} "$DIST/Stirling-PDF-${{ matrix.name }}.AppImage" \;
           fi
 
       # Verify the MSI AND the inner exe extracted from it are signed.

.gitignore

@@ -23,6 +23,10 @@ customFiles/
 configs/
 watchedFolders/
 clientWebUI/
+# Scratch dir used by local fixture-regeneration runs (see
+# app/proprietary/src/test/resources/db-migration-fixtures/README.md).
+# Holds downloaded JARs and disposable workdirs. Never committed.
+.alpha-local/
 !cucumber/
 !cucumber/exampleFiles/
 !cucumber/exampleFiles/example_html.zip
@@ -53,6 +57,8 @@ app/core/src/main/resources/static/robots.txt
 app/core/src/main/resources/static/pdfium/
 app/core/src/main/resources/static/pdfjs/
 app/core/src/main/resources/static/vendor/
+app/core/src/main/resources/static/**/*.gz
+app/core/src/main/resources/static/**/*.br
 # Note: Keep backend-managed files like fonts/, css/, js/, pdfjs/, etc.
 
 # Gradle
@@ -174,7 +180,6 @@ venv.bak/
 
 # Env files (secrets / local overrides). Subproject .gitignore files whitelist any committed defaults.
 .env*
-!.env.saas.example
 
 # VS Code
 /.vscode/**/*
@@ -209,7 +214,7 @@ out/
 *.asc
 
 # Allow test fixture certificates (synthetic, no real credentials)
-!frontend/src/core/tests/test-fixtures/certs/**
+!frontend/editor/src/core/tests/test-fixtures/certs/**
 
 # SSH Keys
 *.pub
@@ -251,7 +256,7 @@ node_modules/
 *compact*.json
 test_batch.json
 *.backup.*.json
-frontend/public/locales/*/translation.backup*.json
+frontend/editor/public/locales/*/translation.backup*.json
 
 # Development/build artifacts
 .gradle-cache/
@@ -274,3 +279,6 @@ docs/type3/signatures/
 # Playwright MCP screenshots / traces
 .playwright-mcp/
 *.playwright-mcp.png
+
+# Local screenshot artifacts from *-screenshots.spec.ts
+frontend/editor/screenshots/

.gitleaksignore

@@ -0,0 +1,5 @@
+# PostHog project-level key — phc_ prefix keys are public/client-side by design
+# (PostHog client-side tracking embeds them in the browser bundle). Committed
+# intentionally in #6150 so engine/.env has a working default, with real
+# credentials overridden via engine/.env.local.
+engine/.env:generic-api-key:41

.taskfiles/backend.yml

@@ -22,12 +22,13 @@ tasks:
     vars:
       PORT: '{{.PORT | default "8080"}}'
       AIENGINE_URL: '{{.AIENGINE_URL | default ""}}'
+      AIENGINE_TIMEOUTSECONDS: '{{.AIENGINE_TIMEOUTSECONDS | default "120"}}'
     env:
       SERVER_PORT: '{{.PORT}}'
     cmds:
-      - cmd: '{{if .AIENGINE_URL}}AIENGINE_URL={{.AIENGINE_URL}} AIENGINE_ENABLED=true {{end}}cmd /c ".\gradlew.bat :stirling-pdf:bootRun"'
+      - cmd: '{{if .AIENGINE_URL}}AIENGINE_URL={{.AIENGINE_URL}} AIENGINE_ENABLED=true AIENGINE_TIMEOUTSECONDS={{.AIENGINE_TIMEOUTSECONDS}} {{end}}cmd /c ".\gradlew.bat :stirling-pdf:bootRun"'
         platforms: [windows]
-      - cmd: '{{if .AIENGINE_URL}}AIENGINE_URL={{.AIENGINE_URL}} AIENGINE_ENABLED=true {{end}}./gradlew :stirling-pdf:bootRun'
+      - cmd: '{{if .AIENGINE_URL}}AIENGINE_URL={{.AIENGINE_URL}} AIENGINE_ENABLED=true AIENGINE_TIMEOUTSECONDS={{.AIENGINE_TIMEOUTSECONDS}} {{end}}./gradlew :stirling-pdf:bootRun'
         platforms: [linux, darwin]
 
   dev:bundled:
@@ -40,10 +41,10 @@ tasks:
         platforms: [linux, darwin]
 
   dev:saas:
-    desc: "Start backend in SaaS flavor against Supabase (loads .env.saas.local)"
+    desc: "Start backend in SaaS flavor against Supabase"
     # `dotenv:` reads from the root Taskfile's directory (".") because this
-    # subtaskfile is included with `dir: .`. Drop the file at the repo root.
-    dotenv: ['.env.saas.local']
+    # subtaskfile is included with `dir: .`.
+    dotenv: ['app/.env.saas.local', 'app/.env.saas']
     ignore_error: true
     vars:
       PORT: '{{.PORT | default "8080"}}'

.taskfiles/desktop.yml

@@ -78,6 +78,13 @@ tasks:
     cmds:
       - npx tauri build --bundles appimage
 
+  test:
+    desc: "Run Tauri/Cargo tests"
+    deps: [prepare]
+    dir: editor/src-tauri
+    cmds:
+      - cargo test
+
   clean:
     desc: "Clean Tauri/Cargo build artifacts"
     dir: editor
@@ -126,8 +133,29 @@ tasks:
         --no-header-files
         --no-man-pages
         --output runtime/jre
+      # jlink emits its files mode 444 (read-only). Tauri's build-script
+      # resource copier preserves source permissions when staging
+      # `runtime/jre/**/*` into `target/<profile>/runtime/jre/...`, so the
+      # staged copies are read-only too. On any subsequent incremental
+      # build the copier tries to overwrite them and fails with a bare
+      # `Permission denied (os error 13)` (Rust's io::Error Display drops
+      # the path, so the failure is opaque). Make the source writable here
+      # so the staged destinations are writable and can be overwritten.
+      #
+      # Trade-off: this task runs for both `task desktop:dev` and
+      # `task desktop:build`, so production bundles also ship mode-644
+      # JRE files instead of 444. Functionally harmless on POSIX (the
+      # `other` bit is `r--` either way, and on macOS code signing is the
+      # real integrity check) and on Windows the DOS read-only attribute
+      # isn't load-bearing for the bundled JDK. If we ever need strict
+      # 444 in production, split the chmod into a dev-only step and have
+      # `desktop:build` run `jlink:clean` first to force a fresh build.
+      - cmd: chmod -R u+w runtime/jre
+        platforms: [linux, darwin]
+      - cmd: powershell -NoProfile -Command "Get-ChildItem -Recurse runtime/jre | ForEach-Object { $_.IsReadOnly = $false }"
+        platforms: [windows]
     status:
-      - test -d editor/src-tauri/runtime/jre
+      - test -f runtime/jre/release
 
   jlink:clean:
     desc: "Remove JLink runtime and bundled JARs"

.taskfiles/e2e.yml

@@ -34,6 +34,9 @@ tasks:
     ignore_error: true
     vars:
       BASE_DIR: '{{.ROOT_DIR}}/.test-state/playwright'
+    # COVERAGE=1 in the calling environment attaches the JaCoCo agent to
+    # the bootRun JVM and writes to BASE_DIR/jacoco.exec on shutdown.
+    # Off by default to keep local dev runs uninstrumented; CI flips it.
     env:
       STIRLING_BASE_PATH: '{{.BASE_DIR}}'
       # Suppress the analytics opt-in modal that fires on first admin login.
@@ -58,12 +61,19 @@ tasks:
           set -e
           rm -rf "{{.BASE_DIR}}"
           mkdir -p "{{.BASE_DIR}}"
+          GRADLE_ARGS=":stirling-pdf:bootRun"
+          if [ -n "${COVERAGE:-}" ]; then
+            # copyJacocoAgent is wired as a dependency of bootRun when
+            # -PjacocoAgent=true, so we do not need to invoke it separately.
+            GRADLE_ARGS="$GRADLE_ARGS -PjacocoAgent=true -PjacocoExec={{.BASE_DIR}}/jacoco.exec"
+            echo "JaCoCo coverage enabled, writing to {{.BASE_DIR}}/jacoco.exec"
+          fi
           # Background gradle and record its PID so the runner can clean up
           # the exact process tree (wrapper + forked Spring Boot JVM) without
           # resorting to fuzzy `pkill -f` patterns. `wait` keeps this script
           # alive for the lifetime of gradle so Task'"'"'s parallel deps stay
           # synchronised.
-          bash gradlew :stirling-pdf:bootRun > "{{.BASE_DIR}}/backend.log" 2>&1 &
+          bash gradlew $GRADLE_ARGS > "{{.BASE_DIR}}/backend.log" 2>&1 &
           GRADLE_PID=$!
           echo $GRADLE_PID > "{{.BASE_DIR}}/backend.pid"
           wait $GRADLE_PID

.taskfiles/frontend.yml

@@ -112,6 +112,12 @@ tasks:
       - task: dev:_run
         vars: { MODE: prototypes, PORT: '{{.PORT}}', BACKEND_URL: '{{.BACKEND_URL}}', OPEN: '{{.OPEN}}' }
 
+  dev:portal:
+    desc: "Start developer portal dev server"
+    deps: [install]
+    cmds:
+      - npx vite portal --port {{.PORT | default "5173"}}{{if .OPEN}} --open{{end}}
+
   # ============================================================
   # Build
   # ============================================================
@@ -156,6 +162,24 @@ tasks:
     cmds:
       - npx vite build editor --mode prototypes
 
+  build:portal:
+    desc: "Build developer portal"
+    deps: [install]
+    cmds:
+      - npx vite build portal
+
+  storybook:
+    desc: "Start Storybook dev server"
+    deps: [install]
+    cmds:
+      - npx storybook dev -p 6006 {{.CLI_ARGS}}
+
+  storybook:build:
+    desc: "Build static Storybook"
+    deps: [install]
+    cmds:
+      - npx storybook build {{.CLI_ARGS}}
+
   # ============================================================
   # Code quality
   # ============================================================
@@ -163,9 +187,23 @@ tasks:
   lint:
     desc: "Run linting"
     deps: [install]
+    cmds:
+      - task: lint:eslint
+      - task: lint:dpdm
+
+  lint:eslint:
+    desc: "Run ESLint linting"
+    deps: [install]
     cmds:
       - npx eslint --max-warnings=0
-      - npx dpdm editor/src --circular --no-warning --no-tree --exit-code circular:1
+
+  lint:dpdm:
+    desc: "Run circular import linting"
+    deps: [install]
+    cmds:
+      # Globs so dpdm walks the whole tree. dpdm expands the braces itself, so this is
+      # shell-agnostic. Covers editor, portal, and the shared design system.
+      - npx dpdm "editor/src/**/*.{ts,tsx}" "portal/src/**/*.{ts,tsx}" "shared/**/*.{ts,tsx}" --circular --no-warning --no-tree --exit-code circular:1
 
   lint:fix:
     desc: "Auto-fix lint issues"
@@ -236,6 +274,18 @@ tasks:
     cmds:
       - npx tsc --noEmit --project editor/src/prototypes/tsconfig.json
 
+  typecheck:portal:
+    desc: "Typecheck developer portal build variant"
+    deps: [install]
+    cmds:
+      - npx tsc --noEmit --project portal/tsconfig.json
+
+  typecheck:shared:
+    desc: "Typecheck the shared design system"
+    deps: [install]
+    cmds:
+      - npx tsc --noEmit --project shared/tsconfig.json
+
   typecheck:all:
     desc: "Typecheck all build variants"
     cmds:
@@ -245,6 +295,8 @@ tasks:
       - task: typecheck:desktop
       - task: typecheck:scripts
       - task: typecheck:prototypes
+      - task: typecheck:portal
+      - task: typecheck:shared
 
   # ============================================================
   # Quality Gate
@@ -265,7 +317,9 @@ tasks:
       - task: lint
       - task: format:check
       - task: build
+      - task: build:portal
       - task: test
+      - task: storybook:build
 
   # ============================================================
   # Test
@@ -284,10 +338,25 @@ tasks:
       - npx vitest --watch --root editor
 
   test:coverage:
-    desc: "Run tests with coverage"
+    desc: "Run tests with coverage (one-shot; CI-friendly)."
     deps: [install]
     cmds:
-      - npx vitest --coverage --root editor
+      # `vitest run` makes this CI-safe (the bare `vitest` form enters watch
+      # mode). Explicit reporter list because v8 + json-summary is what the
+      # coverage-summary.py helper consumes; html/text are kept for humans.
+      #
+      # reportsDirectory is pinned to ./coverage relative to vitest's root
+      # (--root editor), so output lands at frontend/editor/coverage/. The
+      # CI upload step reads from that path. An earlier attempt with
+      # `./editor/coverage` double-nested into frontend/editor/editor/coverage;
+      # pinning future-proofs against vitest changing the default.
+      - >
+        npx vitest run --root editor --coverage
+        --coverage.provider=v8
+        --coverage.reporter=text-summary
+        --coverage.reporter=json-summary
+        --coverage.reporter=html
+        --coverage.reportsDirectory=./coverage
 
   # ============================================================
   # Code Generation

LICENSE

@@ -10,14 +10,16 @@ if that directory exists, is licensed under the license defined in "app/propriet
 if that directory exists, is licensed under the license defined in "app/saas/LICENSE".
 * All content that resides under the "engine/" directory of this repository,
 if that directory exists, is licensed under the license defined in "engine/LICENSE".
-* All content that resides under the "frontend/src/proprietary/" directory of this repository,
-if that directory exists, is licensed under the license defined in "frontend/src/proprietary/LICENSE".
-* All content that resides under the "frontend/src/desktop/" directory of this repository,
-if that directory exists, is licensed under the license defined in "frontend/src/desktop/LICENSE".
-* All content that resides under the "frontend/src/saas/" directory of this repository,
-if that directory exists, is licensed under the license defined in "frontend/src/saas/LICENSE".
-* All content that resides under the "frontend/src/prototypes/" directory of this repository,
-if that directory exists, is licensed under the license defined in "frontend/src/prototypes/LICENSE".
+* All content that resides under the "frontend/editor/src/proprietary/" directory of this repository,
+if that directory exists, is licensed under the license defined in "frontend/editor/src/proprietary/LICENSE".
+* All content that resides under the "frontend/editor/src/desktop/" directory of this repository,
+if that directory exists, is licensed under the license defined in "frontend/editor/src/desktop/LICENSE".
+* All content that resides under the "frontend/editor/src/saas/" directory of this repository,
+if that directory exists, is licensed under the license defined in "frontend/editor/src/saas/LICENSE".
+* All content that resides under the "frontend/editor/src/prototypes/" directory of this repository,
+if that directory exists, is licensed under the license defined in "frontend/editor/src/prototypes/LICENSE".
+* All content that resides under the "frontend/portal/" directory of this repository,
+if that directory exists, is licensed under the license defined in "frontend/portal/LICENSE".
 * Content outside of the above mentioned directories or restrictions above is
 available under the MIT License as defined below.
 

Taskfile.yml

@@ -58,6 +58,23 @@ tasks:
           BACKEND_URL: 'http://localhost:{{.BACKEND_PORT}}'
           OPEN: "true"
 
+  dev:saas:
+    desc: "Start SaaS backend + frontend concurrently on free ports"
+    vars:
+      PORTS:
+        sh: '{{if eq OS "windows"}}{{.FIND_FREE_PORT_PS}} 8080 5173{{else}}{{.FIND_FREE_PORT_SH}} 8080 5173{{end}}'
+      BACKEND_PORT: '{{index (splitList "\n" .PORTS) 0}}'
+      FRONTEND_PORT: '{{index (splitList "\n" .PORTS) 1}}'
+    deps:
+      - task: backend:dev:saas
+        vars:
+          PORT: '{{.BACKEND_PORT}}'
+      - task: frontend:dev:saas
+        vars:
+          PORT: '{{.FRONTEND_PORT}}'
+          BACKEND_URL: 'http://localhost:{{.BACKEND_PORT}}'
+          OPEN: "true"
+
   dev:all:
     desc: "Start backend + frontend + engine concurrently on free ports"
     vars:
@@ -74,7 +91,7 @@ tasks:
         vars:
           PORT: '{{.BACKEND_PORT}}'
           AIENGINE_URL: 'http://localhost:{{.ENGINE_PORT}}'
-      - task: frontend:dev:prototypes
+      - task: frontend:dev
         vars:
           PORT: '{{.FRONTEND_PORT}}'
           BACKEND_URL: 'http://localhost:{{.BACKEND_PORT}}'

app/.env.saas

@@ -1,20 +1,20 @@
 ###############################################################################
-# Stirling-PDF SaaS local environment template.
+# Stirling-PDF SaaS environment defaults.
 #
-# Copy this file to `.env.saas.local` (gitignored) and fill in real values.
-# Loaded by `task backend:dev:saas` via Taskfile's `dotenv:` directive, then
-# read by Spring Boot's `${...}` placeholders in application-saas.properties
-# and application-dev.properties.
+# This file is committed and provides non-secret defaults loaded by
+# `task backend:dev:saas`. Put real values for secrets (passwords, project
+# refs, edge function secrets) in `.env.saas.local` - any variable set there
+# takes precedence over what's defined here.
 #
-# DO NOT commit `.env.saas.local`. Only `.env.saas.example` is checked in.
+# DO NOT commit `.env.saas.local`. Only `.env.saas` is checked in.
 ###############################################################################
 
 # ---------- Supabase project ----------
 # Project reference (the subdomain part of <ref>.supabase.co). Required.
-# Example dev project:
+# Set in .env.saas.local.
 SAAS_DB_PROJECT_REF=
 
-# Edge function secret used by billing/license rollup calls.
+# Edge function secret used by billing/license rollup calls. Set in .env.saas.local.
 SUPABASE_EDGE_FUNCTION_SECRET=
 
 # ---------- Database (saas profile) ----------
@@ -28,7 +28,7 @@ SAAS_DB_PASSWORD=
 # ---------- Database (dev profile overrides) ----------
 # Used when `--spring.profiles.include=dev` is active. The dev profile
 # defaults the URL/username to the shared dev Supabase project, but the
-# password must still be provided here.
+# password must still be provided in .env.saas.local.
 SAAS_DEV_DB_URL=
 SAAS_DEV_DB_USERNAME=postgres
 SAAS_DEV_DB_PASSWORD=

app/.gitignore

@@ -0,0 +1,3 @@
+# Whitelist committed env defaults. `.env.saas.local` (and any other .env*)
+# stays ignored via the root .gitignore.
+!.env.saas

app/allowed-licenses.json

@@ -44,6 +44,10 @@
       "moduleName": ".*",
       "moduleLicense": "The MIT License"
     },
+    {
+      "moduleName": ".*",
+      "moduleLicense": "MIT-0"
+    },
     {
       "moduleName": "com.github.jai-imageio:jai-imageio-core",
       "moduleLicense": "LICENSE.txt"

app/common/build.gradle

@@ -60,7 +60,7 @@ dependencies {
         exclude group: 'com.google.code.gson', module: 'gson'
     }
 
-    api 'com.stirling:jpdfium:1.0.1'
+    api 'com.stirling:jpdfium:1.0.2'
 
     // -PjpdfiumPlatforms=all|<csv of linux-x64,linux-arm64,darwin-x64,darwin-arm64,windows-x64>
     def jpdfiumPlatformsProp = (project.findProperty('jpdfiumPlatforms') ?: 'all').toString().trim()
@@ -75,9 +75,12 @@ dependencies {
     }
     logger.lifecycle("JPDFium native platforms: ${jpdfiumPlatforms.join(', ')}")
     jpdfiumPlatforms.each { platform ->
-        runtimeOnly "com.stirling:jpdfium-natives-${platform}:1.0.1"
+        runtimeOnly "com.stirling:jpdfium-natives-${platform}:1.0.2"
     }
 
+    // Bucket4j (local in-process token bucket for RateLimitStore default impl)
+    implementation 'com.bucket4j:bucket4j_jdk17-core:8.19.0'
+
     // ArchUnit: enforces module dependency direction (see ArchitectureTest)
     testImplementation 'com.tngtech.archunit:archunit-junit5:1.4.2'
 }

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

@@ -0,0 +1,142 @@
+package stirling.software.SPDF.pdf.parser;
+
+import java.util.ArrayList;
+import java.util.List;
+
+/**
+ * Detects whether a page is one- or two-column from per-line bounding boxes, and classifies an
+ * X-span into the column it belongs to. Detection is a midpoint vote at {@code pageWidth / 2}.
+ *
+ * <p>Capped at two columns by design — sufficient for the redaction target set (single-column
+ * documents and IEEE-style two-column papers). 3+ column layouts (newspapers, magazines) and
+ * off-centre gutters (asymmetric two-column) would need a histogram or clustering approach to
+ * detect the actual gutter X. (future work)
+ *
+ * <p>Coordinates are PDFTextStripper screen space (top-left origin, Y increases downward).
+ */
+public final class PageColumnLayout {
+
+    /**
+     * Slack when checking "crosses a gutter" so single-pixel overshoots don't mark a line as
+     * spanning.
+     */
+    public static final float SPAN_SLACK_PT = 2f;
+
+    /**
+     * Slack on each side of the page midpoint inside which a line is considered "spanning"
+     * (covering both columns) rather than belonging to one side.
+     */
+    private static final float MIDPOINT_SLACK_PT = 30f;
+
+    /**
+     * Minimum line width (points) for a line to count toward the two-column tally. Avoids false
+     * positives where right-aligned dates, page numbers, or short "Link" fragments next to a
+     * heading look like a second column when they're really just inline metadata.
+     */
+    private static final float MIN_COLUMN_LINE_WIDTH_PT = 100f;
+
+    /**
+     * Minimum number of clearly leftish AND clearly rightish lines (each of width &ge; {@link
+     * #MIN_COLUMN_LINE_WIDTH_PT}) required to call the page two-column. Anything below this falls
+     * back to single-column.
+     */
+    private static final int MIN_SIDE_LINES = 3;
+
+    private final List<float[]> columns;
+    private final List<float[]> gutters;
+
+    private PageColumnLayout(List<float[]> columns, List<float[]> gutters) {
+        this.columns = columns;
+        this.gutters = gutters;
+    }
+
+    /**
+     * Determines column layout from per-line bounding boxes ({@code [x1, _, x2, _]}). Counts lines
+     * whose X-midpoint sits clearly left of, or clearly right of, the page midpoint (with {@link
+     * #MIDPOINT_SLACK_PT} slack each side). If both sides have at least {@link #MIN_SIDE_LINES}
+     * lines, the page is treated as two-column with the gutter at the page midpoint. Otherwise it's
+     * single-column.
+     *
+     * <p>Cross-column lines must already be split: callers should feed boxes from a line extractor
+     * that splits same-Y glyphs at large X gaps (see {@code AllTextLineExtractor}). Without that
+     * split, IEEE-style aligned-baseline 2-column PDFs produce one wide merged box per row and the
+     * side tallies all end up classified as "spanning", falling to single-column.
+     */
+    public static PageColumnLayout fromLineBoxes(List<float[]> lineBoxes, float pageWidth) {
+        if (lineBoxes == null || lineBoxes.isEmpty()) {
+            return new PageColumnLayout(List.of(new float[] {0f, pageWidth}), List.of());
+        }
+        float pageMid = pageWidth / 2f;
+        int left = 0, right = 0;
+        for (float[] lb : lineBoxes) {
+            if (lb == null || lb.length < 3) continue;
+            float width = lb[2] - lb[0];
+            // Skip narrow lines — dates, page numbers, "Link" labels next to a heading should
+            // not, on their own, make a single-column doc look two-column.
+            if (width < MIN_COLUMN_LINE_WIDTH_PT) continue;
+            float mid = (lb[0] + lb[2]) * 0.5f;
+            if (mid < pageMid - MIDPOINT_SLACK_PT) left++;
+            else if (mid > pageMid + MIDPOINT_SLACK_PT) right++;
+        }
+        if (left < MIN_SIDE_LINES || right < MIN_SIDE_LINES) {
+            return new PageColumnLayout(List.of(new float[] {0f, pageWidth}), List.of());
+        }
+        float gutterL = pageMid - MIDPOINT_SLACK_PT;
+        float gutterR = pageMid + MIDPOINT_SLACK_PT;
+        return new PageColumnLayout(
+                List.of(new float[] {0f, gutterL}, new float[] {gutterR, pageWidth}),
+                List.of(new float[] {gutterL, gutterR}));
+    }
+
+    /** All columns, left-to-right, as {@code [leftX, rightX]} pairs. Never empty. */
+    public List<float[]> columns() {
+        return columns;
+    }
+
+    /** Gutters between columns, left-to-right, as {@code [leftX, rightX]} pairs. */
+    public List<float[]> gutters() {
+        return gutters;
+    }
+
+    public int columnCount() {
+        return columns.size();
+    }
+
+    /**
+     * Returns the column index containing the X-midpoint of {@code [x1, x2]}, falling back to the
+     * closest column if the midpoint sits inside a gutter.
+     */
+    public int columnOf(float x1, float x2) {
+        float mid = (x1 + x2) * 0.5f;
+        int best = 0;
+        float bestDist = Float.MAX_VALUE;
+        for (int i = 0; i < columns.size(); i++) {
+            float[] c = columns.get(i);
+            if (mid >= c[0] && mid <= c[1]) return i;
+            float dist = mid < c[0] ? c[0] - mid : mid - c[1];
+            if (dist < bestDist) {
+                bestDist = dist;
+                best = i;
+            }
+        }
+        return best;
+    }
+
+    /**
+     * Returns every column index whose X-range overlaps {@code [x1, x2]} with at least {@link
+     * #SPAN_SLACK_PT} of intrusion. A normal in-column line returns one index; a line crossing a
+     * gutter returns two or more.
+     */
+    public int[] columnsCrossing(float x1, float x2) {
+        List<Integer> hits = new ArrayList<>();
+        for (int i = 0; i < columns.size(); i++) {
+            float[] c = columns.get(i);
+            float overlap = Math.min(x2, c[1]) - Math.max(x1, c[0]);
+            if (overlap > SPAN_SLACK_PT) hits.add(i);
+        }
+        if (hits.isEmpty()) hits.add(columnOf(x1, x2));
+        int[] out = new int[hits.size()];
+        for (int i = 0; i < hits.size(); i++) out[i] = hits.get(i);
+        return out;
+    }
+}

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

@@ -0,0 +1,133 @@
+package stirling.software.SPDF.pdf.parser;
+
+import java.awt.geom.Point2D;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+
+import org.apache.pdfbox.contentstream.PDFGraphicsStreamEngine;
+import org.apache.pdfbox.cos.COSName;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.graphics.image.PDImage;
+import org.apache.pdfbox.util.Matrix;
+
+/**
+ * PDFGraphicsStreamEngine that intercepts {@code drawImage} calls and records each image's bounding
+ * box in PDF user-space (origin bottom-left, Y up) by transforming the unit square through the
+ * current transformation matrix (CTM).
+ *
+ * <p>Usage:
+ *
+ * <pre>{@code
+ * PageImageLocator locator = new PageImageLocator(page, pageIndex);
+ * locator.processPage(page);
+ * List<ImageBox> boxes = locator.getImageBoxes();
+ * }</pre>
+ *
+ * <p>Each {@link ImageBox} carries the 0-based page index and the axis-aligned bounding box {@code
+ * (x1, y1, x2, y2)} in PDF user-space coordinates.
+ */
+public final class PageImageLocator extends PDFGraphicsStreamEngine {
+
+    /**
+     * Bounding box of a raster or vector image found on a PDF page.
+     *
+     * @param pageIndex 0-based page index
+     * @param x1 left edge in PDF user-space (origin bottom-left)
+     * @param y1 bottom edge in PDF user-space
+     * @param x2 right edge
+     * @param y2 top edge
+     */
+    public record ImageBox(int pageIndex, float x1, float y1, float x2, float y2) {}
+
+    private final int pageIndex;
+    private final List<ImageBox> imageBoxes = new ArrayList<>();
+    private final Point2D.Float currentPoint = new Point2D.Float();
+
+    /**
+     * @param page the PDPage to process
+     * @param pageIndex 0-based index of this page in the document (stored on each returned {@link
+     *     ImageBox})
+     */
+    public PageImageLocator(PDPage page, int pageIndex) {
+        super(page);
+        this.pageIndex = pageIndex;
+    }
+
+    /** Returns all image bounding boxes collected during {@link #processPage}. */
+    public List<ImageBox> getImageBoxes() {
+        return imageBoxes;
+    }
+
+    @Override
+    public void drawImage(PDImage pdImage) throws IOException {
+        Matrix ctm = getGraphicsState().getCurrentTransformationMatrix();
+        // An image occupies the unit square (0,0)→(1,1) in image space.
+        // Transform all four corners through the CTM to get the page-space bounding box.
+        float a = ctm.getScaleX();
+        float b = ctm.getShearY();
+        float c = ctm.getShearX();
+        float d = ctm.getScaleY();
+        float e = ctm.getTranslateX();
+        float f = ctm.getTranslateY();
+        float[] xs = {e, a + e, c + e, a + c + e};
+        float[] ys = {f, b + f, d + f, b + d + f};
+        float x1 = Float.MAX_VALUE, y1 = Float.MAX_VALUE;
+        float x2 = -Float.MAX_VALUE, y2 = -Float.MAX_VALUE;
+        for (float x : xs) {
+            x1 = Math.min(x1, x);
+            x2 = Math.max(x2, x);
+        }
+        for (float y : ys) {
+            y1 = Math.min(y1, y);
+            y2 = Math.max(y2, y);
+        }
+        imageBoxes.add(new ImageBox(pageIndex, x1, y1, x2, y2));
+    }
+
+    // ---------- required abstract methods (no-op for path operations) ----------
+
+    @Override
+    public void appendRectangle(Point2D p0, Point2D p1, Point2D p2, Point2D p3) {}
+
+    @Override
+    public void clip(int windingRule) {}
+
+    @Override
+    public void moveTo(float x, float y) {
+        currentPoint.setLocation(x, y);
+    }
+
+    @Override
+    public void lineTo(float x, float y) {
+        currentPoint.setLocation(x, y);
+    }
+
+    @Override
+    public void curveTo(float x1, float y1, float x2, float y2, float x3, float y3) {
+        currentPoint.setLocation(x3, y3);
+    }
+
+    @Override
+    public Point2D getCurrentPoint() {
+        return currentPoint;
+    }
+
+    @Override
+    public void closePath() {}
+
+    @Override
+    public void endPath() {}
+
+    @Override
+    public void strokePath() {}
+
+    @Override
+    public void fillPath(int windingRule) {}
+
+    @Override
+    public void fillAndStrokePath(int windingRule) {}
+
+    @Override
+    public void shadingFill(COSName shadingName) {}
+}

app/common/src/main/java/stirling/software/common/annotations/AutoJobPostMapping.java

@@ -77,6 +77,10 @@ public @interface AutoJobPostMapping {
     /**
      * Relative resource weight (1-100). See {@link
      * stirling.software.common.enumeration.ResourceWeight} for the standard tiers.
+     *
+     * <p>The default is a sentinel ({@link Integer#MIN_VALUE}); {@code
+     * AutoJobPostMappingWeightTest} fails the build if any endpoint leaves it unset. Runtime
+     * readers clamp the value into {@code [1, 100]}.
      */
-    int resourceWeight() default 1;
+    int resourceWeight() default Integer.MIN_VALUE;
 }

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

@@ -0,0 +1,24 @@
+package stirling.software.common.cluster;
+
+/** Health and identity facade for the active cluster backplane. */
+public interface ClusterBackplane {
+
+    /** Returns {@code true} when the backplane is reachable; used for health endpoints. */
+    boolean isHealthy();
+
+    /** Returns {@code "inprocess"} or {@code "valkey"}. */
+    String backplaneType();
+
+    /** Returns this JVM's stable node id (matches {@code Cluster.resolvedNodeId()}). */
+    String localNodeId();
+
+    /**
+     * Whether this JVM should run the local {@link
+     * stirling.software.common.service.TaskManager#cleanupOldJobs()} loop. Distributed backplanes
+     * own job expiry via their own TTL, so they should override this to return {@code false}.
+     * Defaults to {@code true} so in-process behavior is preserved without an explicit override.
+     */
+    default boolean shouldRunLocalCleanup() {
+        return true;
+    }
+}

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

@@ -0,0 +1,63 @@
+package stirling.software.common.cluster;
+
+import org.springframework.context.annotation.Configuration;
+
+import jakarta.annotation.PostConstruct;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.model.ApplicationProperties.Cluster;
+
+/**
+ * Validates that cluster mode is internally consistent.
+ *
+ * <p>Cluster settings are bound on the central {@link ApplicationProperties} under {@code
+ * cluster.*}; this class reads {@link ApplicationProperties#getCluster()} and runs guards in {@link
+ * PostConstruct}. When {@code cluster.enabled=false} (the default) all checks are skipped so a
+ * single-instance install needs no new config.
+ */
+@Slf4j
+@Configuration
+@RequiredArgsConstructor
+public class ClusterConfig {
+
+    private final ApplicationProperties applicationProperties;
+
+    @PostConstruct
+    void validate() {
+        Cluster cluster = applicationProperties.getCluster();
+        if (!cluster.isEnabled()) {
+            return;
+        }
+        String backplane = cluster.getBackplane();
+        if ("valkey".equalsIgnoreCase(backplane)) {
+            String url = cluster.getValkey() == null ? null : cluster.getValkey().getUrl();
+            if (url == null || url.isBlank()) {
+                throw new IllegalStateException(
+                        "cluster.enabled=true with backplane=valkey requires"
+                                + " cluster.valkey.url to be set (e.g."
+                                + " redis://valkey:6379).");
+            }
+        } else if ("inprocess".equalsIgnoreCase(backplane)) {
+            // enabled+inprocess only coordinates the local JVM; cross-node lookups will 410.
+            log.warn(
+                    "cluster.enabled=true with backplane=inprocess - only the local"
+                            + " JVM is coordinated. Cross-node lookups and the file proxy will fail."
+                            + " Use backplane=valkey for real multi-node deployments.");
+        } else {
+            // Fail fast on typos like "valky" so Spring doesn't later report a cryptic
+            // "no ClusterBackplane bean" - the operator-facing error names the bad value.
+            throw new IllegalStateException(
+                    "cluster.enabled=true with unknown backplane '"
+                            + backplane
+                            + "'. Valid values: inprocess | valkey.");
+        }
+        log.info(
+                "Cluster mode enabled (backplane={}, nodeRole={}, nodeId={}).",
+                backplane,
+                cluster.resolvedRole(),
+                cluster.resolvedNodeId());
+    }
+}

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

@@ -0,0 +1,12 @@
+package stirling.software.common.cluster;
+
+import java.time.Instant;
+
+/**
+ * Snapshot of a peer node as recorded in the {@link InstanceRegistry}.
+ *
+ * @param internalAddress {@code host:port} the node listens on for {@code /internal/cluster/**}
+ * @param role one of {@code WEB}, {@code WORKER}, {@code BOTH}
+ */
+public record ClusterNode(
+        String nodeId, String internalAddress, Instant lastHeartbeat, String role) {}

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

@@ -0,0 +1,21 @@
+package stirling.software.common.cluster;
+
+import java.time.Duration;
+import java.util.Optional;
+
+/** Cluster-wide mutual exclusion primitive; non-reentrant by contract. */
+public interface DistributedLock {
+
+    Optional<LockHandle> tryAcquire(String lockKey, Duration leaseTime);
+
+    interface LockHandle extends AutoCloseable {
+        void release();
+
+        boolean renew(Duration leaseTime);
+
+        @Override
+        default void close() {
+            release();
+        }
+    }
+}

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

@@ -0,0 +1,45 @@
+package stirling.software.common.cluster;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.file.Files;
+import java.nio.file.Path;
+
+/** Low-level storage seam for result/job files. */
+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 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.
+     */
+    default Stored store(Path source, String originalName) throws IOException {
+        try (InputStream in = Files.newInputStream(source)) {
+            return store(in, originalName);
+        }
+    }
+
+    /** Open the stored file for streaming reads. Caller closes. */
+    InputStream retrieve(String fileId) throws IOException;
+
+    /** Load the stored file into a byte array. */
+    byte[] retrieveBytes(String fileId) throws IOException;
+
+    /** Size of the stored file in bytes. */
+    long size(String fileId) throws IOException;
+
+    /** Delete the stored file. Returns true if a file was removed. */
+    boolean delete(String fileId);
+
+    /** Whether the file id exists in the store. */
+    boolean exists(String fileId);
+}

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

@@ -0,0 +1,18 @@
+package stirling.software.common.cluster;
+
+import java.time.Duration;
+import java.util.Collection;
+import java.util.Optional;
+
+/** Maps {@code nodeId} to its internal cluster address, with TTL'd heartbeats. */
+public interface InstanceRegistry {
+
+    /** Register or refresh this node. Idempotent so a wiped backplane self-heals on next tick. */
+    void register(ClusterNode node, Duration heartbeatTtl);
+
+    Optional<ClusterNode> lookup(String nodeId);
+
+    Collection<ClusterNode> activeNodes();
+
+    void deregister(String nodeId);
+}

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

@@ -0,0 +1,24 @@
+package stirling.software.common.cluster;
+
+import java.time.Duration;
+import java.util.Collection;
+import java.util.Optional;
+
+/** Cluster-visible storage for job status and result metadata, with TTL'd entries. */
+public interface JobStore {
+
+    /** Persist or overwrite a job entry. {@code ttl} sets the lifetime of the entry. */
+    void put(JobStoreEntry entry, Duration ttl);
+
+    Optional<JobStoreEntry> get(String jobId);
+
+    void delete(String jobId);
+
+    boolean exists(String jobId);
+
+    /** Reverse lookup: which job owns this result file id? */
+    Optional<String> findJobIdByFileId(String fileId);
+
+    /** Snapshot of every active entry. Used by admin/stats endpoints; may be O(n). */
+    Collection<JobStoreEntry> all();
+}

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

@@ -0,0 +1,30 @@
+package stirling.software.common.cluster;
+
+import java.time.Instant;
+import java.util.List;
+import java.util.Map;
+
+/**
+ * Cluster-visible projection of a job's status and result metadata, as persisted in {@link
+ * JobStore}.
+ *
+ * @param owningNodeId the node id that originally executed the job
+ */
+public record JobStoreEntry(
+        String jobId,
+        JobState state,
+        String owningNodeId,
+        Instant createdAt,
+        Instant completedAt,
+        String error,
+        List<String> fileIds,
+        Map<String, String> resultMeta) {
+
+    /** Lifecycle states for a job as observed by the cluster. */
+    public enum JobState {
+        PENDING,
+        RUNNING,
+        COMPLETE,
+        FAILED
+    }
+}

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

@@ -0,0 +1,16 @@
+package stirling.software.common.cluster;
+
+import java.time.Duration;
+import java.util.Optional;
+
+/** Short-TTL namespaced key/value cache backed by the cluster backplane. */
+public interface KeyValueCache {
+
+    void put(String namespace, String key, String value, Duration ttl);
+
+    Optional<String> get(String namespace, String key);
+
+    void evict(String namespace, String key);
+
+    void evictNamespace(String namespace);
+}

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

@@ -0,0 +1,24 @@
+package stirling.software.common.cluster;
+
+import java.time.Duration;
+
+/**
+ * Token-bucket rate limiting backed by the cluster backplane.
+ *
+ * <p>In-process implementations enforce a per-JVM limit; distributed implementations enforce a
+ * single global limit across every node. Both use a Bucket4j greedy-refill token bucket so the
+ * semantics match across single-node and cluster deployments.
+ */
+public interface RateLimitStore {
+
+    /**
+     * Attempt to consume one token from the bucket identified by {@code bucketKey}.
+     *
+     * @param bucketKey opaque key identifying the bucket (e.g. {@code api:user:123})
+     * @param capacity bucket capacity
+     * @param refillPeriod time window over which {@code capacity} tokens refill
+     */
+    RateLimitDecision tryConsume(String bucketKey, long capacity, Duration refillPeriod);
+
+    record RateLimitDecision(boolean allowed, long remainingTokens, long nanosToWaitForRefill) {}
+}

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

@@ -0,0 +1,7 @@
+package stirling.software.common.cluster;
+
+/** Records one increment per sticky-session miss (a 410 Gone for a job owned by another node). */
+@FunctionalInterface
+public interface StickyMissRecorder {
+    void recordStickyMiss();
+}

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

@@ -0,0 +1,31 @@
+package stirling.software.common.cluster.inprocess;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.cluster.ClusterBackplane;
+import stirling.software.common.model.ApplicationProperties;
+
+@Slf4j
+public class InProcessClusterBackplane implements ClusterBackplane {
+
+    private final ApplicationProperties applicationProperties;
+
+    public InProcessClusterBackplane(ApplicationProperties applicationProperties) {
+        this.applicationProperties = applicationProperties;
+    }
+
+    @Override
+    public boolean isHealthy() {
+        return true;
+    }
+
+    @Override
+    public String backplaneType() {
+        return "inprocess";
+    }
+
+    @Override
+    public String localNodeId() {
+        return applicationProperties.getCluster().resolvedNodeId();
+    }
+}

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

@@ -0,0 +1,65 @@
+package stirling.software.common.cluster.inprocess;
+
+import org.springframework.boot.autoconfigure.condition.ConditionalOnExpression;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnMissingBean;
+import org.springframework.context.annotation.Bean;
+import org.springframework.context.annotation.Configuration;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.cluster.ClusterBackplane;
+import stirling.software.common.cluster.DistributedLock;
+import stirling.software.common.cluster.InstanceRegistry;
+import stirling.software.common.cluster.JobStore;
+import stirling.software.common.cluster.KeyValueCache;
+import stirling.software.common.cluster.RateLimitStore;
+import stirling.software.common.model.ApplicationProperties;
+
+/**
+ * Default cluster backplane wiring: every interface gets an {@code InProcess*} bean. Active when
+ * cluster mode is off or {@code cluster.backplane=inprocess}.
+ */
+@Slf4j
+@Configuration
+@ConditionalOnExpression(
+        "!${cluster.enabled:false} ||"
+                + " '${cluster.backplane:inprocess}'.equalsIgnoreCase('inprocess')")
+public class InProcessClusterConfiguration {
+
+    @Bean
+    @ConditionalOnMissingBean
+    public ClusterBackplane clusterBackplane(ApplicationProperties applicationProperties) {
+        log.info("Cluster backplane: in-process (single node)");
+        return new InProcessClusterBackplane(applicationProperties);
+    }
+
+    @Bean
+    @ConditionalOnMissingBean
+    public JobStore jobStore() {
+        return new InProcessJobStore();
+    }
+
+    @Bean
+    @ConditionalOnMissingBean
+    public RateLimitStore rateLimitStore() {
+        return new InProcessRateLimitStore();
+    }
+
+    @Bean
+    @ConditionalOnMissingBean
+    public DistributedLock distributedLock() {
+        return new InProcessDistributedLock();
+    }
+
+    @Bean
+    @ConditionalOnMissingBean
+    public KeyValueCache keyValueCache() {
+        return new InProcessKeyValueCache();
+    }
+
+    @Bean
+    @ConditionalOnMissingBean
+    public InstanceRegistry instanceRegistry() {
+        return new InProcessInstanceRegistry();
+    }
+}

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

@@ -0,0 +1,128 @@
+package stirling.software.common.cluster.inprocess;
+
+import java.time.Duration;
+import java.util.Optional;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.concurrent.atomic.AtomicLong;
+
+import stirling.software.common.cluster.DistributedLock;
+
+/**
+ * In-process {@link DistributedLock}, non-reentrant per the interface contract, with lease-expiry
+ * semantics that mirror a SET-NX-EX style distributed backend.
+ *
+ * <p>Each lock state carries a per-acquire {@code ownerToken} and an {@code expiryNanos}; another
+ * caller can take over once the lease has elapsed even if the original holder never called {@link
+ * LockHandle#release()}. This matters mostly for parity with the Valkey-backed implementation
+ * (Redis {@code SETEX} auto-expires the key); within a single JVM a crashed holder takes its lock
+ * state with it, but tests and code that rely on the {@code leaseTime} parameter still need it to
+ * be honored.
+ *
+ * <p>Expiry is lazy: an expired lock state lingers in the map until the next acquire attempt for
+ * the same key replaces it. Per-key cleanup also happens on explicit {@link LockHandle#release()},
+ * so a balanced acquire/release workload keeps the map size bounded.
+ */
+public class InProcessDistributedLock implements DistributedLock {
+
+    private final ConcurrentHashMap<String, LockState> locks = new ConcurrentHashMap<>();
+    private final AtomicLong tokenSeq = new AtomicLong();
+
+    /**
+     * Lease state for a single acquired lock. {@code ownerToken} prevents a former holder from
+     * releasing or renewing a lock now owned by someone else after lease expiry; {@code
+     * expiryNanos} is read/written only inside {@link ConcurrentHashMap#compute} so the bin lock
+     * provides the necessary happens-before guarantee.
+     */
+    private static final class LockState {
+        final long ownerToken;
+        long expiryNanos;
+
+        LockState(long ownerToken, long expiryNanos) {
+            this.ownerToken = ownerToken;
+            this.expiryNanos = expiryNanos;
+        }
+    }
+
+    @Override
+    public Optional<LockHandle> tryAcquire(String lockKey, Duration leaseTime) {
+        long token = tokenSeq.incrementAndGet();
+        long nowNanos = System.nanoTime();
+        long expiryNanos = nowNanos + leaseTime.toNanos();
+        boolean[] acquired = {false};
+        locks.compute(
+                lockKey,
+                (k, existing) -> {
+                    if (existing == null || existing.expiryNanos - nowNanos <= 0L) {
+                        // No lock, or the previous lease has expired - we take it. Subtraction
+                        // form avoids the long-overflow trap that would bite a naive
+                        // expiryNanos <= nowNanos comparison around System.nanoTime() rollover.
+                        acquired[0] = true;
+                        return new LockState(token, expiryNanos);
+                    }
+                    return existing;
+                });
+        if (!acquired[0]) {
+            return Optional.empty();
+        }
+        return Optional.of(new InProcessHandle(lockKey, token));
+    }
+
+    private void releaseInternal(String lockKey, long token) {
+        locks.compute(
+                lockKey,
+                (k, existing) -> {
+                    if (existing == null || existing.ownerToken != token) {
+                        // Already removed, expired-and-replaced, or never ours.
+                        return existing;
+                    }
+                    return null;
+                });
+    }
+
+    private boolean renewInternal(String lockKey, long token, Duration leaseTime) {
+        long nowNanos = System.nanoTime();
+        boolean[] renewed = {false};
+        locks.compute(
+                lockKey,
+                (k, existing) -> {
+                    if (existing == null
+                            || existing.ownerToken != token
+                            || existing.expiryNanos - nowNanos <= 0L) {
+                        // Lock is gone or expired; renewal is a no-op so the caller can detect it.
+                        return existing;
+                    }
+                    existing.expiryNanos = nowNanos + leaseTime.toNanos();
+                    renewed[0] = true;
+                    return existing;
+                });
+        return renewed[0];
+    }
+
+    private final class InProcessHandle implements LockHandle {
+        private final String lockKey;
+        private final long token;
+        private boolean released;
+
+        InProcessHandle(String lockKey, long token) {
+            this.lockKey = lockKey;
+            this.token = token;
+        }
+
+        @Override
+        public synchronized void release() {
+            if (released) {
+                return;
+            }
+            released = true;
+            releaseInternal(lockKey, token);
+        }
+
+        @Override
+        public synchronized boolean renew(Duration leaseTime) {
+            if (released) {
+                return false;
+            }
+            return renewInternal(lockKey, token, leaseTime);
+        }
+    }
+}

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

@@ -0,0 +1,42 @@
+package stirling.software.common.cluster.inprocess;
+
+import java.time.Duration;
+import java.util.Collection;
+import java.util.Collections;
+import java.util.Optional;
+import java.util.concurrent.atomic.AtomicReference;
+
+import stirling.software.common.cluster.ClusterNode;
+import stirling.software.common.cluster.InstanceRegistry;
+
+public class InProcessInstanceRegistry implements InstanceRegistry {
+
+    private final AtomicReference<ClusterNode> self = new AtomicReference<>();
+
+    @Override
+    public void register(ClusterNode node, Duration heartbeatTtl) {
+        self.set(node);
+    }
+
+    @Override
+    public Optional<ClusterNode> lookup(String nodeId) {
+        ClusterNode current = self.get();
+        return current != null && current.nodeId().equals(nodeId)
+                ? Optional.of(current)
+                : Optional.empty();
+    }
+
+    @Override
+    public Collection<ClusterNode> activeNodes() {
+        ClusterNode current = self.get();
+        return current == null ? Collections.emptyList() : Collections.singletonList(current);
+    }
+
+    @Override
+    public void deregister(String nodeId) {
+        ClusterNode current = self.get();
+        if (current != null && current.nodeId().equals(nodeId)) {
+            self.set(null);
+        }
+    }
+}

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

@@ -0,0 +1,96 @@
+package stirling.software.common.cluster.inprocess;
+
+import java.time.Duration;
+import java.time.Instant;
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.List;
+import java.util.Map;
+import java.util.Optional;
+import java.util.concurrent.ConcurrentHashMap;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.cluster.JobStore;
+import stirling.software.common.cluster.JobStoreEntry;
+
+@Slf4j
+public class InProcessJobStore implements JobStore {
+
+    private final ConcurrentHashMap<String, Holder> entries = new ConcurrentHashMap<>();
+
+    @Override
+    public void put(JobStoreEntry entry, Duration ttl) {
+        Instant expiry = ttl == null ? Instant.MAX : Instant.now().plus(ttl);
+        entries.put(entry.jobId(), new Holder(entry, expiry));
+    }
+
+    @Override
+    public Optional<JobStoreEntry> get(String jobId) {
+        Holder h = entries.get(jobId);
+        if (h == null) {
+            return Optional.empty();
+        }
+        if (h.isExpired()) {
+            entries.remove(jobId, h);
+            return Optional.empty();
+        }
+        return Optional.of(h.entry);
+    }
+
+    @Override
+    public void delete(String jobId) {
+        entries.remove(jobId);
+    }
+
+    @Override
+    public boolean exists(String jobId) {
+        return get(jobId).isPresent();
+    }
+
+    @Override
+    public Optional<String> findJobIdByFileId(String fileId) {
+        for (Map.Entry<String, Holder> e : entries.entrySet()) {
+            Holder h = e.getValue();
+            if (h.isExpired()) {
+                continue;
+            }
+            List<String> fileIds = h.entry.fileIds();
+            if (fileIds != null && fileIds.contains(fileId)) {
+                return Optional.of(e.getKey());
+            }
+        }
+        return Optional.empty();
+    }
+
+    @Override
+    public Collection<JobStoreEntry> all() {
+        List<JobStoreEntry> result = new ArrayList<>(entries.size());
+        for (Holder h : entries.values()) {
+            if (!h.isExpired()) {
+                result.add(h.entry);
+            }
+        }
+        return result;
+    }
+
+    /** Drop entries whose TTL has elapsed. Called by the {@code TaskManager} cleanup scheduler. */
+    public int purgeExpired() {
+        int removed = 0;
+        Instant now = Instant.now();
+        for (Map.Entry<String, Holder> e : entries.entrySet()) {
+            if (!e.getValue().expiry.equals(Instant.MAX) && e.getValue().expiry.isBefore(now)) {
+                if (entries.remove(e.getKey(), e.getValue())) {
+                    removed++;
+                }
+            }
+        }
+        return removed;
+    }
+
+    private record Holder(JobStoreEntry entry, Instant expiry) {
+        boolean isExpired() {
+            return !expiry.equals(Instant.MAX) && expiry.isBefore(Instant.now());
+        }
+    }
+}

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

@@ -0,0 +1,55 @@
+package stirling.software.common.cluster.inprocess;
+
+import java.time.Duration;
+import java.time.Instant;
+import java.util.Map;
+import java.util.Optional;
+import java.util.concurrent.ConcurrentHashMap;
+
+import stirling.software.common.cluster.KeyValueCache;
+
+public class InProcessKeyValueCache implements KeyValueCache {
+
+    private final ConcurrentHashMap<String, ConcurrentHashMap<String, Expiring>> namespaces =
+            new ConcurrentHashMap<>();
+
+    @Override
+    public void put(String namespace, String key, String value, Duration ttl) {
+        Instant expiry = ttl == null ? Instant.MAX : Instant.now().plus(ttl);
+        namespaces
+                .computeIfAbsent(namespace, n -> new ConcurrentHashMap<>())
+                .put(key, new Expiring(value, expiry));
+    }
+
+    @Override
+    public Optional<String> get(String namespace, String key) {
+        Map<String, Expiring> ns = namespaces.get(namespace);
+        if (ns == null) {
+            return Optional.empty();
+        }
+        Expiring e = ns.get(key);
+        if (e == null) {
+            return Optional.empty();
+        }
+        if (e.expiry.isBefore(Instant.now())) {
+            ns.remove(key, e);
+            return Optional.empty();
+        }
+        return Optional.of(e.value);
+    }
+
+    @Override
+    public void evict(String namespace, String key) {
+        Map<String, Expiring> ns = namespaces.get(namespace);
+        if (ns != null) {
+            ns.remove(key);
+        }
+    }
+
+    @Override
+    public void evictNamespace(String namespace) {
+        namespaces.remove(namespace);
+    }
+
+    private record Expiring(String value, Instant expiry) {}
+}

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

@@ -0,0 +1,49 @@
+package stirling.software.common.cluster.inprocess;
+
+import java.time.Duration;
+import java.util.Collections;
+import java.util.LinkedHashMap;
+import java.util.Map;
+
+import io.github.bucket4j.Bandwidth;
+import io.github.bucket4j.Bucket;
+import io.github.bucket4j.ConsumptionProbe;
+import io.github.bucket4j.local.LocalBucketBuilder;
+
+import stirling.software.common.cluster.RateLimitStore;
+
+/** Bucket4j-backed token bucket implementation of {@link RateLimitStore}. */
+public class InProcessRateLimitStore implements RateLimitStore {
+
+    /** Cap to bound memory; oldest accessed buckets are evicted. */
+    private static final int MAX_BUCKETS = 10_000;
+
+    private final Map<String, Bucket> buckets =
+            Collections.synchronizedMap(
+                    new LinkedHashMap<String, Bucket>(256, 0.75f, true) {
+                        @Override
+                        protected boolean removeEldestEntry(Map.Entry<String, Bucket> eldest) {
+                            return size() > MAX_BUCKETS;
+                        }
+                    });
+
+    @Override
+    public RateLimitDecision tryConsume(String bucketKey, long capacity, Duration refillPeriod) {
+        String compositeKey = bucketKey + "|" + capacity + "|" + refillPeriod.toNanos();
+        Bucket bucket =
+                buckets.computeIfAbsent(compositeKey, k -> buildBucket(capacity, refillPeriod));
+        ConsumptionProbe probe = bucket.tryConsumeAndReturnRemaining(1);
+        return new RateLimitDecision(
+                probe.isConsumed(),
+                probe.getRemainingTokens(),
+                probe.isConsumed() ? 0L : probe.getNanosToWaitForRefill());
+    }
+
+    private static Bucket buildBucket(long capacity, Duration refillPeriod) {
+        Bandwidth limit =
+                Bandwidth.builder().capacity(capacity).refillGreedy(capacity, refillPeriod).build();
+        LocalBucketBuilder builder = Bucket.builder();
+        builder.addLimit(limit);
+        return builder.build();
+    }
+}

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

@@ -0,0 +1,128 @@
+package stirling.software.common.cluster.inprocess;
+
+import java.io.BufferedInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.util.UUID;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.common.cluster.FileStore;
+
+/** Local-disk {@link FileStore} storing files under a base directory keyed by a UUID file id. */
+@Slf4j
+public class LocalDiskFileStore implements FileStore {
+
+    private final String baseDirPath;
+
+    public LocalDiskFileStore(String baseDirPath) {
+        this.baseDirPath = baseDirPath;
+    }
+
+    @Override
+    public Stored store(InputStream in, String originalName) throws IOException {
+        String fileId = UUID.randomUUID().toString();
+        Path filePath = resolve(fileId);
+        Files.createDirectories(filePath.getParent());
+        boolean success = false;
+        try {
+            long size = Files.copy(in, filePath);
+            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);
+                }
+            }
+        }
+    }
+
+    /**
+     * File-to-file copy. {@link Files#copy(Path, Path, java.nio.file.CopyOption...)} can use {@code
+     * sendfile(2)} on Linux for a zero-copy kernel transfer when source and destination share a
+     * filesystem, avoiding the streaming overhead of pulling the bytes through the JVM heap. Reads
+     * the source size before copying so the post-copy stat is unnecessary.
+     */
+    @Override
+    public Stored store(Path source, String originalName) throws IOException {
+        String fileId = UUID.randomUUID().toString();
+        Path filePath = resolve(fileId);
+        Files.createDirectories(filePath.getParent());
+        long size = Files.size(source);
+        boolean success = false;
+        try {
+            Files.copy(source, filePath);
+            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);
+                }
+            }
+        }
+    }
+
+    @Override
+    public InputStream retrieve(String fileId) throws IOException {
+        return new BufferedInputStream(Files.newInputStream(resolve(fileId)));
+    }
+
+    @Override
+    public byte[] retrieveBytes(String fileId) throws IOException {
+        Path filePath = resolve(fileId);
+        if (!Files.exists(filePath)) {
+            throw new IOException("File not found with ID: " + fileId);
+        }
+        return Files.readAllBytes(filePath);
+    }
+
+    @Override
+    public long size(String fileId) throws IOException {
+        Path filePath = resolve(fileId);
+        if (!Files.exists(filePath)) {
+            throw new IOException("File not found with ID: " + fileId);
+        }
+        return Files.size(filePath);
+    }
+
+    @Override
+    public boolean delete(String fileId) {
+        try {
+            return Files.deleteIfExists(resolve(fileId));
+        } catch (IOException e) {
+            log.error("Error deleting file with ID: {}", fileId, e);
+            return false;
+        }
+    }
+
+    @Override
+    public boolean exists(String fileId) {
+        return Files.exists(resolve(fileId));
+    }
+
+    public Path resolve(String fileId) {
+        if (fileId.contains("..") || fileId.contains("/") || fileId.contains("\\")) {
+            throw new IllegalArgumentException("Invalid file ID");
+        }
+        Path basePath = Path.of(baseDirPath).normalize().toAbsolutePath();
+        Path resolvedPath = basePath.resolve(fileId).normalize();
+        if (!resolvedPath.startsWith(basePath)) {
+            throw new IllegalArgumentException("File ID resolves to an invalid path");
+        }
+        return resolvedPath;
+    }
+}

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

@@ -0,0 +1,29 @@
+package stirling.software.common.cluster.inprocess;
+
+import org.springframework.beans.factory.annotation.Value;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnMissingBean;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.context.annotation.Bean;
+import org.springframework.context.annotation.Configuration;
+
+import stirling.software.common.cluster.FileStore;
+
+/**
+ * Always-on wiring for the per-node local-disk {@link FileStore}. Active when {@code
+ * cluster.artifactStore=local} (the default; {@code matchIfMissing=true}). The S3 artifact-store
+ * supplies its own bean when {@code cluster.artifactStore=s3}.
+ */
+@Configuration
+@ConditionalOnProperty(
+        prefix = "cluster",
+        name = "artifactStore",
+        havingValue = "local",
+        matchIfMissing = true)
+public class LocalDiskFileStoreConfiguration {
+
+    @Bean
+    @ConditionalOnMissingBean
+    public FileStore fileStore(@Value("${stirling.tempDir:/tmp/stirling-files}") String tempDir) {
+        return new LocalDiskFileStore(tempDir);
+    }
+}

app/common/src/main/java/stirling/software/common/configuration/ConfigInitializer.java

@@ -80,6 +80,7 @@ public class ConfigInitializer {
             YamlHelper settingsFile = new YamlHelper(settingTempPath);
 
             migrateEnterpriseEditionToPremium(settingsFile, settingsTemplateFile);
+            migrateProFeaturesKeyCasing(settingsFile, settingsTemplateFile);
 
             boolean changesMade =
                     settingsTemplateFile.updateValuesFromYaml(settingsFile, settingsTemplateFile);
@@ -116,31 +117,52 @@ public class ConfigInitializer {
         }
         if (yaml.getValueByExactKeyPath("enterpriseEdition", "SSOAutoLogin") != null) {
             template.updateValue(
-                    List.of("premium", "proFeatures", "SSOAutoLogin"),
+                    List.of("premium", "proFeatures", "ssoAutoLogin"),
                     yaml.getValueByExactKeyPath("enterpriseEdition", "SSOAutoLogin"));
         }
         if (yaml.getValueByExactKeyPath("enterpriseEdition", "CustomMetadata", "autoUpdateMetadata")
                 != null) {
             template.updateValue(
-                    List.of("premium", "proFeatures", "CustomMetadata", "autoUpdateMetadata"),
+                    List.of("premium", "proFeatures", "customMetadata", "autoUpdateMetadata"),
                     yaml.getValueByExactKeyPath(
                             "enterpriseEdition", "CustomMetadata", "autoUpdateMetadata"));
         }
         if (yaml.getValueByExactKeyPath("enterpriseEdition", "CustomMetadata", "author") != null) {
             template.updateValue(
-                    List.of("premium", "proFeatures", "CustomMetadata", "author"),
+                    List.of("premium", "proFeatures", "customMetadata", "author"),
                     yaml.getValueByExactKeyPath("enterpriseEdition", "CustomMetadata", "author"));
         }
         if (yaml.getValueByExactKeyPath("enterpriseEdition", "CustomMetadata", "creator") != null) {
             template.updateValue(
-                    List.of("premium", "proFeatures", "CustomMetadata", "creator"),
+                    List.of("premium", "proFeatures", "customMetadata", "creator"),
                     yaml.getValueByExactKeyPath("enterpriseEdition", "CustomMetadata", "creator"));
         }
         if (yaml.getValueByExactKeyPath("enterpriseEdition", "CustomMetadata", "producer")
                 != null) {
             template.updateValue(
-                    List.of("premium", "proFeatures", "CustomMetadata", "producer"),
+                    List.of("premium", "proFeatures", "customMetadata", "producer"),
                     yaml.getValueByExactKeyPath("enterpriseEdition", "CustomMetadata", "producer"));
         }
     }
+
+    // TODO: Remove post migration
+    // settings.yml.template renamed the two non-camelCase proFeatures keys
+    // ("SSOAutoLogin" -> "ssoAutoLogin", "CustomMetadata" -> "customMetadata") so the whole
+    // settings pipeline is consistent camelCase. The save path (YamlHelper.updateValue) matches
+    // keys case-sensitively, so without this carry-forward an existing install's values written
+    // under the old PascalCase keys would be dropped on upgrade and reset to template defaults.
+    void migrateProFeaturesKeyCasing(YamlHelper yaml, YamlHelper template) {
+        Object ssoAutoLogin = yaml.getValueByExactKeyPath("premium", "proFeatures", "SSOAutoLogin");
+        if (ssoAutoLogin != null) {
+            template.updateValue(List.of("premium", "proFeatures", "ssoAutoLogin"), ssoAutoLogin);
+        }
+        for (String field : List.of("autoUpdateMetadata", "author", "creator", "producer")) {
+            Object value =
+                    yaml.getValueByExactKeyPath("premium", "proFeatures", "CustomMetadata", field);
+            if (value != null) {
+                template.updateValue(
+                        List.of("premium", "proFeatures", "customMetadata", field), value);
+            }
+        }
+    }
 }

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

@@ -13,6 +13,7 @@ import java.util.Arrays;
 import java.util.Collection;
 import java.util.List;
 import java.util.Locale;
+import java.util.UUID;
 
 import org.springframework.boot.context.properties.ConfigurationProperties;
 import org.springframework.context.annotation.Bean;
@@ -77,6 +78,7 @@ public class ApplicationProperties {
     private PdfEditor pdfEditor = new PdfEditor();
     private AiEngine aiEngine = new AiEngine();
     private InternalApi internalApi = new InternalApi();
+    private Cluster cluster = new Cluster();
 
     @Bean
     public PropertySource<?> dynamicYamlPropertySource(ConfigurableEnvironment environment)
@@ -254,6 +256,106 @@ public class ApplicationProperties {
         private int longRunningTimeoutSeconds = 600;
     }
 
+    /**
+     * 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
+     * off; when off the in-process backplane is wired and no other cluster keys are required.
+     */
+    @Data
+    public static class Cluster {
+
+        /** Master switch. When {@code false} (default) the in-process backplane is wired. */
+        private boolean enabled = false;
+
+        /** Backplane implementation selector. Valid values: {@code inprocess} | {@code valkey}. */
+        private String backplane = "inprocess";
+
+        /**
+         * Transient cluster job-artifact store selector. Valid values: {@code local} | {@code s3}.
+         *
+         * <p>This is distinct from {@code storage.provider}, which selects the backend for
+         * persistent user-uploaded files. The two switches exist because the user-facing storage
+         * feature is optional ({@code storage.enabled=false} is common) but every multi-node
+         * cluster still needs a shared artifact store to serve cross-node downloads. Both
+         * implementations share credentials from {@code storage.s3.*} when set to {@code s3}.
+         */
+        private String artifactStore = "local";
+
+        private Valkey valkey = new Valkey();
+        private Node node = new Node();
+
+        private transient String cachedNodeId;
+
+        public NodeRole resolvedRole() {
+            if (node == null || node.getRole() == null) {
+                return NodeRole.BOTH;
+            }
+            String value = node.getRole().trim().toUpperCase(Locale.ROOT);
+            try {
+                return NodeRole.valueOf(value);
+            } catch (IllegalArgumentException ex) {
+                return NodeRole.BOTH;
+            }
+        }
+
+        public synchronized String resolvedNodeId() {
+            if (node != null && node.getId() != null && !node.getId().isBlank()) {
+                return node.getId();
+            }
+            if (cachedNodeId == null) {
+                cachedNodeId = UUID.randomUUID().toString();
+            }
+            return cachedNodeId;
+        }
+
+        public enum NodeRole {
+            WEB,
+            WORKER,
+            BOTH
+        }
+
+        @Data
+        public static class Valkey {
+            /**
+             * {@code redis://host:6379} or {@code rediss://...} for TLS. Required when cluster mode
+             * is on and backplane is valkey.
+             */
+            private String url = "";
+
+            private Tls tls = new Tls();
+
+            @Data
+            public static class Tls {
+                /**
+                 * When {@code true}, skip Valkey/Redis TLS certificate verification (dev/test
+                 * only). Leave {@code false} in production.
+                 */
+                private boolean skipCertVerification = false;
+            }
+        }
+
+        @Data
+        public static class Node {
+            /** Optional explicit node id. Blank = auto-generated UUID at startup. */
+            private String id = "";
+
+            /** {@code web} | {@code worker} | {@code both}. */
+            private String role = "both";
+
+            /**
+             * Internal cluster address advertised in the instance registry (host:port). Blank =
+             * derived at startup.
+             */
+            private String internalAddress = "";
+
+            /** {@code http} | {@code https} - scheme used when peers call this node. */
+            private String scheme = "http";
+
+            /** Heartbeat publish interval for the instance registry, in milliseconds. */
+            private long heartbeatIntervalMs = 5000;
+        }
+    }
+
     /**
      * HTTP timeouts for loopback calls to internal Stirling API endpoints, used by the AI workflow
      * executor and the pipeline processor. A bounded read timeout prevents a hung tool (e.g. an
@@ -426,6 +528,16 @@ public class ApplicationProperties {
             private String provider;
             private Client client = new Client();
 
+            /**
+             * When true, the OAuth2/OIDC login flow logs the full set of ID token and UserInfo
+             * claims at INFO level (and again at ERROR level if the username attribute cannot be
+             * resolved). Used to diagnose provider misconfiguration (for example ADFS not returning
+             * an {@code email} claim). WARNING: writes PII (sub, email, name) to application logs.
+             * Leave disabled in production; enable only while actively troubleshooting and disable
+             * again afterwards.
+             */
+            private Boolean debugLogging = false;
+
             public void setScopes(String scopes) {
                 List<String> scopesList =
                         Arrays.stream(scopes.split(",")).map(String::trim).toList();
@@ -676,6 +788,7 @@ public class ApplicationProperties {
         private boolean enabled = false;
         private String provider = "local";
         private Local local = new Local();
+        private S3 s3 = new S3();
         private Quotas quotas = new Quotas();
         private Sharing sharing = new Sharing();
         private Signing signing = new Signing();
@@ -685,6 +798,57 @@ public class ApplicationProperties {
             private String basePath = InstallationPathConfig.getPath() + "storage";
         }
 
+        @Data
+        public static class S3 {
+            /**
+             * Optional custom endpoint (e.g. {@code https://<account>.r2.cloudflarestorage.com},
+             * {@code https://<project>.supabase.co/storage/v1/s3}, or {@code http://localhost:9000}
+             * for MinIO). Blank = use AWS regional default.
+             */
+            private String endpoint = "";
+
+            private String bucket = "";
+
+            private String region = "us-east-1";
+
+            private String accessKey = "";
+            private String secretKey = "";
+
+            /**
+             * When {@code true} use path-style URLs ({@code <endpoint>/<bucket>/<key>}) instead of
+             * virtual-hosted ({@code <bucket>.<endpoint>/<key>}). MinIO and most S3-compatible
+             * gateways require path-style; AWS S3 prefers virtual-hosted.
+             */
+            private boolean pathStyleAccess = false;
+
+            /**
+             * When {@code false} (default), {@code endpoint} hostnames that resolve to private,
+             * loopback, or link-local addresses are rejected at startup to block SSRF attacks via
+             * the cloud metadata service (e.g. {@code http://169.254.169.254/}). Set to {@code
+             * true} to opt in for MinIO / in-cluster S3 endpoints on private networks.
+             */
+            private boolean allowPrivateEndpoints = false;
+
+            /**
+             * Controls when the SDK adds an {@code x-amz-checksum-*} header on PUT/UploadPart.
+             * Default {@code WHEN_SUPPORTED} (the SDK default since 2.30) makes the SDK send a
+             * CRC32 checksum on every upload - this works on AWS S3, MinIO, current Supabase,
+             * Backblaze B2 (post-July-2025), and modern R2. Set to {@code WHEN_REQUIRED} to
+             * suppress the auto-checksum on vendors that reject unknown {@code x-amz-checksum-*}
+             * headers (older Backblaze B2, some R2 corner cases, GCS S3 endpoint). Invalid values
+             * fall back to {@code WHEN_SUPPORTED}.
+             */
+            private String requestChecksumCalculation = "WHEN_SUPPORTED";
+
+            /**
+             * Controls when the SDK validates returned {@code x-amz-checksum-*} headers on GET
+             * responses. Default {@code WHEN_SUPPORTED}. Set to {@code WHEN_REQUIRED} if your
+             * vendor never returns these headers and you see false-positive checksum-mismatch
+             * errors. Invalid values fall back to {@code WHEN_SUPPORTED}.
+             */
+            private String responseChecksumValidation = "WHEN_SUPPORTED";
+        }
+
         @Data
         public static class Sharing {
             private boolean enabled = false;

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

@@ -1,15 +1,13 @@
 package stirling.software.common.service;
 
-import java.io.BufferedInputStream;
-import java.io.BufferedOutputStream;
+import java.io.ByteArrayInputStream;
 import java.io.IOException;
 import java.io.InputStream;
-import java.io.OutputStream;
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.util.UUID;
+import java.io.PipedInputStream;
+import java.io.PipedOutputStream;
+import java.util.concurrent.Executors;
+import java.util.concurrent.atomic.AtomicReference;
 
-import org.springframework.beans.factory.annotation.Value;
 import org.springframework.core.io.Resource;
 import org.springframework.stereotype.Service;
 import org.springframework.web.multipart.MultipartFile;
@@ -18,9 +16,11 @@ import org.springframework.web.servlet.mvc.method.annotation.StreamingResponseBo
 import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 
+import stirling.software.common.cluster.FileStore;
+
 /**
  * Service for storing and retrieving files with unique file IDs. Used by the AutoJobPostMapping
- * system to handle file references.
+ * system to handle file references. Disk I/O is delegated to the injected {@link FileStore} bean.
  */
 @Service
 @RequiredArgsConstructor
@@ -30,251 +30,150 @@ public class FileStorage {
     /** Holds the result of a stream-to-disk store operation: the file ID and the bytes written. */
     public record StoredFile(String fileId, long size) {}
 
-    @Value("${stirling.tempDir:/tmp/stirling-files}")
-    private String tempDirPath;
-
     private final FileOrUploadService fileOrUploadService;
+    private final FileStore fileStore;
 
-    /**
-     * Store a file and return its unique ID
-     *
-     * @param file The file to store
-     * @return The unique ID assigned to the file
-     * @throws IOException If there is an error storing the file
-     */
     public String storeFile(MultipartFile file) throws IOException {
-        String fileId = generateFileId();
-        Path filePath = getFilePath(fileId);
-
-        // Ensure the directory exists
-        Files.createDirectories(filePath.getParent());
-
-        // Transfer the file to the storage location
-        file.transferTo(filePath.toFile());
-
-        log.debug("Stored file with ID: {}", fileId);
-        return fileId;
-    }
-
-    /**
-     * Store a byte array as a file and return its unique ID
-     *
-     * @param bytes The byte array to store
-     * @param originalName The original name of the file (for extension)
-     * @return The unique ID assigned to the file
-     * @throws IOException If there is an error storing the file
-     */
-    public String storeBytes(byte[] bytes, String originalName) throws IOException {
-        String fileId = generateFileId();
-        Path filePath = getFilePath(fileId);
-
-        // Ensure the directory exists
-        Files.createDirectories(filePath.getParent());
-
-        // Write the bytes to the file
-        Files.write(filePath, bytes);
-
-        log.debug("Stored byte array with ID: {}", fileId);
-        return fileId;
-    }
-
-    /**
-     * Retrieve a file by its ID as a MultipartFile
-     *
-     * @param fileId The ID of the file to retrieve
-     * @return The file as a MultipartFile
-     * @throws IOException If the file doesn't exist or can't be read
-     */
-    public MultipartFile retrieveFile(String fileId) throws IOException {
-        Path filePath = getFilePath(fileId);
-
-        if (!Files.exists(filePath)) {
-            throw new IOException("File not found with ID: " + fileId);
+        // 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
+        // the bytes through an 8K buffer. Falls back to the InputStream path for in-memory
+        // multiparts, exotic Resource impls, and anything that does not back onto a File.
+        Resource res;
+        try {
+            res = file.getResource();
+        } catch (RuntimeException ignored) {
+            res = null;
         }
+        if (res != null && res.isFile()) {
+            try {
+                FileStore.Stored stored =
+                        fileStore.store(res.getFile().toPath(), file.getOriginalFilename());
+                log.debug("Stored file with ID: {} (fast path)", stored.fileId());
+                return stored.fileId();
+            } catch (IOException ex) {
+                // Some Resource impls advertise isFile()=true but throw on getFile(); fall through.
+                log.debug("Resource fast path failed, falling back to stream copy", ex);
+            }
+        }
+        try (InputStream in = file.getInputStream()) {
+            FileStore.Stored stored = fileStore.store(in, file.getOriginalFilename());
+            log.debug("Stored file with ID: {}", stored.fileId());
+            return stored.fileId();
+        }
+    }
 
-        byte[] fileData = Files.readAllBytes(filePath);
+    public String storeBytes(byte[] bytes, String originalName) throws IOException {
+        FileStore.Stored stored = fileStore.store(new ByteArrayInputStream(bytes), originalName);
+        log.debug("Stored byte array with ID: {}", stored.fileId());
+        return stored.fileId();
+    }
+
+    public MultipartFile retrieveFile(String fileId) throws IOException {
+        byte[] fileData = fileStore.retrieveBytes(fileId);
         return fileOrUploadService.toMockMultipartFile(fileId, fileData);
     }
 
-    /**
-     * Retrieve a file by its ID as a byte array
-     *
-     * @param fileId The ID of the file to retrieve
-     * @return The file as a byte array
-     * @throws IOException If the file doesn't exist or can't be read
-     */
     public byte[] retrieveBytes(String fileId) throws IOException {
-        Path filePath = getFilePath(fileId);
-
-        if (!Files.exists(filePath)) {
-            throw new IOException("File not found with ID: " + fileId);
-        }
-
-        return Files.readAllBytes(filePath);
+        return fileStore.retrieveBytes(fileId);
     }
 
-    /**
-     * Retrieve a file by its ID as a streaming InputStream. The caller is responsible for closing
-     * the returned stream.
-     *
-     * @param fileId The ID of the file to retrieve
-     * @return A buffered InputStream for the file
-     * @throws IOException If the file doesn't exist or can't be read
-     */
     public InputStream retrieveInputStream(String fileId) throws IOException {
-        Path filePath = getFilePath(fileId);
-        // Let Files.newInputStream throw NoSuchFileException naturally — avoids TOCTOU race
-        // between exists-check and open when another thread may delete concurrently.
-        return new BufferedInputStream(Files.newInputStream(filePath));
+        return fileStore.retrieve(fileId);
     }
 
-    /**
-     * Store data from an InputStream as a file and return its unique ID and byte count. Streams
-     * directly to disk without buffering the entire content in heap.
-     *
-     * @param inputStream The input stream to read from
-     * @param originalName The original name of the file (unused, kept for API symmetry)
-     * @return A {@link StoredFile} containing the file ID and the number of bytes written
-     * @throws IOException If there is an error storing the file
-     */
     public StoredFile storeInputStream(InputStream inputStream, String originalName)
             throws IOException {
-        String fileId = generateFileId();
-        Path filePath = getFilePath(fileId);
-        Files.createDirectories(filePath.getParent());
-        long size = Files.copy(inputStream, filePath);
-        log.debug("Stored input stream with ID: {}", fileId);
-        return new StoredFile(fileId, size);
+        FileStore.Stored stored = fileStore.store(inputStream, originalName);
+        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 fileId = generateFileId();
-        Path filePath = getFilePath(fileId);
-        Files.createDirectories(filePath.getParent());
-        boolean success = false;
-        try (OutputStream os = new BufferedOutputStream(Files.newOutputStream(filePath))) {
-            body.writeTo(os);
-            success = true;
-        } finally {
-            if (!success) {
+        // 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.
+        AtomicReference<Throwable> bodyError = new AtomicReference<>();
+        try (PipedOutputStream out = new PipedOutputStream();
+                PipedInputStream in = new PipedInputStream(out, 8192)) {
+            var executor = Executors.newSingleThreadExecutor(Thread.ofVirtual().factory());
+            java.util.concurrent.Future<?> task = null;
+            try {
+                task =
+                        executor.submit(
+                                () -> {
+                                    try {
+                                        body.writeTo(out);
+                                    } catch (Throwable ex) {
+                                        bodyError.set(ex);
+                                    } finally {
+                                        try {
+                                            out.close();
+                                        } catch (IOException ignored) {
+                                            // closed on the consumer side too
+                                        }
+                                    }
+                                });
+                FileStore.Stored stored = fileStore.store(in, originalName);
+                Throwable writerErr = bodyError.get();
+                if (writerErr != null) {
+                    // Body failed mid-write: the FileStore persisted a truncated entry.
+                    // Best-effort delete so we don't leak partial files; never let cleanup
+                    // mask the original writer error.
+                    try {
+                        fileStore.delete(stored.fileId());
+                    } catch (RuntimeException cleanupEx) {
+                        log.warn(
+                                "Failed to delete partial file {} after writer error: {}",
+                                stored.fileId(),
+                                cleanupEx.getMessage());
+                    }
+                    if (writerErr instanceof IOException ioe) {
+                        throw ioe;
+                    }
+                    throw new IOException(
+                            "StreamingResponseBody writer failed: " + writerErr.getMessage(),
+                            writerErr);
+                }
+                log.debug("Stored StreamingResponseBody with ID: {}", stored.fileId());
+                return stored.fileId();
+            } finally {
+                // Interrupt and join the writer task: shutdown() alone returns immediately and a
+                // failed store leaves the writer running, leaking a thread per failed upload.
+                if (task != null && !task.isDone()) {
+                    task.cancel(true);
+                }
+                executor.shutdown();
                 try {
-                    Files.deleteIfExists(filePath);
-                } catch (IOException cleanupEx) {
-                    log.warn(
-                            "Failed to clean up partial file {} after store failure",
-                            filePath,
-                            cleanupEx);
+                    if (!executor.awaitTermination(5, java.util.concurrent.TimeUnit.SECONDS)) {
+                        executor.shutdownNow();
+                    }
+                } catch (InterruptedException ie) {
+                    executor.shutdownNow();
+                    Thread.currentThread().interrupt();
                 }
             }
         }
-        log.debug("Stored StreamingResponseBody with ID: {}", fileId);
-        return fileId;
     }
 
-    /**
-     * Persist a {@link Resource} body to disk, returning the generated file ID. Used by the async
-     * job pipeline to capture {@code ResponseEntity<Resource>} results produced by controllers.
-     */
     public String storeFromResource(Resource resource, String originalName) throws IOException {
-        String fileId = generateFileId();
-        Path filePath = getFilePath(fileId);
-        Files.createDirectories(filePath.getParent());
-        boolean success = false;
         try (InputStream in = resource.getInputStream()) {
-            Files.copy(in, filePath);
-            success = true;
-        } finally {
-            if (!success) {
-                try {
-                    Files.deleteIfExists(filePath);
-                } catch (IOException cleanupEx) {
-                    log.warn(
-                            "Failed to clean up partial file {} after store failure",
-                            filePath,
-                            cleanupEx);
-                }
-            }
+            FileStore.Stored stored = fileStore.store(in, originalName);
+            log.debug("Stored Resource with ID: {}", stored.fileId());
+            return stored.fileId();
         }
-        log.debug("Stored Resource with ID: {}", fileId);
-        return fileId;
     }
 
-    /**
-     * Delete a file by its ID
-     *
-     * @param fileId The ID of the file to delete
-     * @return true if the file was deleted, false otherwise
-     */
     public boolean deleteFile(String fileId) {
-        try {
-            Path filePath = getFilePath(fileId);
-            return Files.deleteIfExists(filePath);
-        } catch (IOException e) {
-            log.error("Error deleting file with ID: {}", fileId, e);
-            return false;
-        }
+        return fileStore.delete(fileId);
     }
 
-    /**
-     * Check if a file exists by its ID
-     *
-     * @param fileId The ID of the file to check
-     * @return true if the file exists, false otherwise
-     */
     public boolean fileExists(String fileId) {
-        Path filePath = getFilePath(fileId);
-        return Files.exists(filePath);
+        return fileStore.exists(fileId);
     }
 
-    /**
-     * Get the size of a file by its ID without loading the content into memory
-     *
-     * @param fileId The ID of the file
-     * @return The size of the file in bytes
-     * @throws IOException If the file doesn't exist or can't be read
-     */
     public long getFileSize(String fileId) throws IOException {
-        Path filePath = getFilePath(fileId);
-
-        if (!Files.exists(filePath)) {
-            throw new IOException("File not found with ID: " + fileId);
-        }
-
-        return Files.size(filePath);
-    }
-
-    /**
-     * Get the path for a file ID
-     *
-     * @param fileId The ID of the file
-     * @return The path to the file
-     * @throws IllegalArgumentException if fileId contains path traversal characters or resolves
-     *     outside base directory
-     */
-    private Path getFilePath(String fileId) {
-        // Validate fileId to prevent path traversal
-        if (fileId.contains("..") || fileId.contains("/") || fileId.contains("\\")) {
-            throw new IllegalArgumentException("Invalid file ID");
-        }
-
-        Path basePath = Path.of(tempDirPath).normalize().toAbsolutePath();
-        Path resolvedPath = basePath.resolve(fileId).normalize();
-
-        // Ensure resolved path is within the base directory
-        if (!resolvedPath.startsWith(basePath)) {
-            throw new IllegalArgumentException("File ID resolves to an invalid path");
-        }
-
-        return resolvedPath;
-    }
-
-    /**
-     * Generate a unique file ID
-     *
-     * @return A unique file ID
-     */
-    private String generateFileId() {
-        return UUID.randomUUID().toString();
+        return fileStore.size(fileId);
     }
 }

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

@@ -57,85 +57,38 @@ public class JobExecutorService {
         this.resourceMonitor = resourceMonitor;
         this.jobQueue = jobQueue;
 
-        // Parse session timeout and calculate effective timeout once during initialization
         long sessionTimeoutMs = parseSessionTimeout(sessionTimeout);
         this.effectiveTimeoutMs = Math.min(asyncRequestTimeoutMs, sessionTimeoutMs);
         log.debug(
                 "Job executor configured with effective timeout of {} ms", this.effectiveTimeoutMs);
     }
 
-    /**
-     * Run a job either asynchronously or synchronously
-     *
-     * @param async Whether to run the job asynchronously
-     * @param work The work to be done
-     * @return The response
-     */
     public ResponseEntity<?> runJobGeneric(boolean async, Supplier<Object> work) {
         return runJobGeneric(async, work, -1);
     }
 
-    /**
-     * Run a job either asynchronously or synchronously with a custom timeout
-     *
-     * @param async Whether to run the job asynchronously
-     * @param work The work to be done
-     * @param customTimeoutMs Custom timeout in milliseconds, or -1 to use the default
-     * @return The response
-     */
     public ResponseEntity<?> runJobGeneric(
             boolean async, Supplier<Object> work, long customTimeoutMs) {
         return runJobGeneric(async, work, customTimeoutMs, false, 50);
     }
 
-    /**
-     * Run a job either asynchronously or synchronously with custom parameters
-     *
-     * @param async Whether to run the job asynchronously
-     * @param work The work to be done
-     * @param customTimeoutMs Custom timeout in milliseconds, or -1 to use the default
-     * @param queueable Whether this job can be queued when system resources are limited
-     * @param resourceWeight The resource weight of this job (1-100)
-     * @return The response
-     */
     public ResponseEntity<?> runJobGeneric(
             boolean async,
             Supplier<Object> work,
             long customTimeoutMs,
             boolean queueable,
             int resourceWeight) {
-        // Generate base UUID
         String baseJobId = UUID.randomUUID().toString();
-
-        // Scope job to authenticated user if security is enabled
         String scopedJobKey = getScopedJobKey(baseJobId);
 
         log.debug("Generated jobId: {} (base: {})", scopedJobKey, baseJobId);
 
-        // Store the scoped job ID in the request for potential use by other components
         if (request != null) {
             request.setAttribute("jobId", scopedJobKey);
-
-            // Also track this job ID in the user's session for authorization purposes
-            // This ensures users can only cancel their own jobs
-            if (request.getSession() != null) {
-                @SuppressWarnings("unchecked")
-                java.util.Set<String> userJobIds =
-                        (java.util.Set<String>) request.getSession().getAttribute("userJobIds");
-
-                if (userJobIds == null) {
-                    userJobIds = new java.util.concurrent.ConcurrentSkipListSet<>();
-                    request.getSession().setAttribute("userJobIds", userJobIds);
-                }
-
-                userJobIds.add(scopedJobKey);
-                log.debug("Added scoped job ID {} to user session", scopedJobKey);
-            }
         }
 
         String jobId = scopedJobKey;
 
-        // Determine which timeout to use
         long timeoutToUse = customTimeoutMs > 0 ? customTimeoutMs : effectiveTimeoutMs;
 
         log.debug(
@@ -146,7 +99,6 @@ public class JobExecutorService {
                 queueable,
                 resourceWeight);
 
-        // Check if we need to queue this job based on resource availability
         boolean shouldQueue =
                 queueable
                         && async
@@ -154,7 +106,6 @@ public class JobExecutorService {
                         resourceMonitor.shouldQueueJob(resourceWeight);
 
         if (shouldQueue) {
-            // Queue the job instead of executing immediately
             log.debug(
                     "Queueing job {} due to resource constraints (weight: {})",
                     jobId,
@@ -162,18 +113,12 @@ public class JobExecutorService {
 
             taskManager.createTask(jobId);
 
-            // Create a specialized wrapper that updates the TaskManager
             final String capturedJobIdForQueue = jobId;
             Supplier<Object> wrappedWork =
                     () -> {
                         try {
-                            // Set jobId in ThreadLocal context for the queued job
                             stirling.software.common.util.JobContext.setJobId(
                                     capturedJobIdForQueue);
-                            log.debug(
-                                    "Set jobId {} in JobContext for queued job execution",
-                                    capturedJobIdForQueue);
-
                             Object result = work.get();
                             processJobResult(capturedJobIdForQueue, result);
                             return result;
@@ -186,21 +131,17 @@ public class JobExecutorService {
                             taskManager.setError(capturedJobIdForQueue, e.getMessage());
                             throw e;
                         } finally {
-                            // Clean up ThreadLocal to avoid memory leaks
                             stirling.software.common.util.JobContext.clear();
                         }
                     };
 
-            // Queue the job and get the future
             CompletableFuture<ResponseEntity<?>> future =
                     jobQueue.queueJob(jobId, resourceWeight, wrappedWork, timeoutToUse);
 
-            // Return immediately with job ID
             return ResponseEntity.ok().body(new JobResponse<>(true, jobId, null));
         } else if (async) {
             taskManager.createTask(jobId);
 
-            // Capture the jobId for the async thread
             final String capturedJobId = jobId;
 
             executor.execute(
@@ -211,13 +152,7 @@ public class JobExecutorService {
                                     capturedJobId,
                                     timeoutToUse);
 
-                            // Set jobId in ThreadLocal context for the async thread
                             stirling.software.common.util.JobContext.setJobId(capturedJobId);
-                            log.debug(
-                                    "Set jobId {} in JobContext for async execution",
-                                    capturedJobId);
-
-                            // Execute with timeout
                             Object result = executeWithTimeout(() -> work.get(), timeoutToUse);
                             processJobResult(capturedJobId, result);
                         } catch (TimeoutException te) {
@@ -227,7 +162,6 @@ public class JobExecutorService {
                             log.error("Error executing job {}: {}", jobId, e.getMessage(), e);
                             taskManager.setError(jobId, e.getMessage());
                         } finally {
-                            // Clean up ThreadLocal to avoid memory leaks
                             stirling.software.common.util.JobContext.clear();
                         }
                     });
@@ -237,27 +171,19 @@ public class JobExecutorService {
             try {
                 log.debug("Running sync job with timeout {} ms", timeoutToUse);
 
-                // Make jobId available to downstream components on the worker thread
                 stirling.software.common.util.JobContext.setJobId(jobId);
-                log.debug("Set jobId {} in JobContext for sync execution", jobId);
-
-                // Execute with timeout
                 Object result = executeWithTimeout(() -> work.get(), timeoutToUse);
 
-                // If the result is already a ResponseEntity, return it directly
                 if (result instanceof ResponseEntity) {
                     return (ResponseEntity<?>) result;
                 }
 
-                // Process different result types
                 return handleResultForSyncJob(result);
             } catch (TimeoutException te) {
                 log.error("Synchronous job timed out after {} ms", timeoutToUse);
                 return ResponseEntity.internalServerError()
                         .body(Map.of("error", "Job timed out after " + timeoutToUse + " ms"));
             } catch (RuntimeException e) {
-                // Check if this is a typed exception that should be handled by
-                // GlobalExceptionHandler (either directly or wrapped)
                 Throwable cause = e.getCause();
                 if (e instanceof IllegalArgumentException
                         || cause
@@ -267,16 +193,13 @@ public class JobExecutorService {
                                 instanceof
                                 stirling.software.common.util.ExceptionUtils
                                         .BaseValidationException) {
-                    // Rethrow so GlobalExceptionHandler can handle with proper HTTP status codes
                     throw e;
                 }
-                // Handle other RuntimeExceptions as generic errors
                 log.error("Error executing synchronous job: {}", e.getMessage(), e);
                 return ResponseEntity.internalServerError()
                         .body(Map.of("error", "Job failed: " + e.getMessage()));
             } catch (Exception e) {
                 log.error("Error executing synchronous job: {}", e.getMessage(), e);
-                // Construct a JSON error response
                 return ResponseEntity.internalServerError()
                         .body(Map.of("error", "Job failed: " + e.getMessage()));
             } finally {
@@ -285,23 +208,13 @@ public class JobExecutorService {
         }
     }
 
-    /**
-     * Process the result of an asynchronous job
-     *
-     * @param jobId The job ID
-     * @param result The result
-     */
     private void processJobResult(String jobId, Object result) {
         try {
             if (result instanceof byte[]) {
-                // Store byte array directly to disk to avoid double memory consumption
                 String fileId = fileStorage.storeBytes((byte[]) result, "result.pdf");
                 taskManager.setFileResult(
                         jobId, fileId, "result.pdf", MediaType.APPLICATION_PDF_VALUE);
                 log.debug("Stored byte[] result with fileId: {}", fileId);
-
-                // Let the byte array get collected naturally in the next GC cycle
-                // We don't need to force System.gc() which can be harmful
             } else if (result instanceof ResponseEntity) {
                 ResponseEntity<?> response = (ResponseEntity<?>) result;
                 Object body = response.getBody();
@@ -330,16 +243,13 @@ public class JobExecutorService {
                     taskManager.setFileResult(jobId, fileId, filename, contentType);
                     log.debug("Stored ResponseEntity<Resource> result with fileId: {}", fileId);
                 } else {
-                    // Check if the response body contains a fileId
                     if (body != null && body.toString().contains("fileId")) {
                         try {
-                            // Try to extract fileId using reflection
                             java.lang.reflect.Method getFileId =
                                     body.getClass().getMethod("getFileId");
                             String fileId = (String) getFileId.invoke(body);
 
                             if (fileId != null && !fileId.isEmpty()) {
-                                // Try to get filename and content type
                                 String filename = "result.pdf";
                                 String contentType = MediaType.APPLICATION_PDF_VALUE;
 
@@ -379,7 +289,6 @@ public class JobExecutorService {
                         }
                     }
 
-                    // Store generic result
                     taskManager.setResult(jobId, body);
                 }
             } else if (result instanceof MultipartFile file) {
@@ -388,16 +297,13 @@ public class JobExecutorService {
                         jobId, fileId, file.getOriginalFilename(), file.getContentType());
                 log.debug("Stored MultipartFile result with fileId: {}", fileId);
             } else {
-                // Check if result has a fileId field
                 if (result != null) {
                     try {
-                        // Try to extract fileId using reflection
                         java.lang.reflect.Method getFileId =
                                 result.getClass().getMethod("getFileId");
                         String fileId = (String) getFileId.invoke(result);
 
                         if (fileId != null && !fileId.isEmpty()) {
-                            // Try to get filename and content type
                             String filename = "result.pdf";
                             String contentType = MediaType.APPLICATION_PDF_VALUE;
 
@@ -435,7 +341,6 @@ public class JobExecutorService {
                     }
                 }
 
-                // Default case: store the result as is
                 taskManager.setResult(jobId, result);
             }
 
@@ -446,16 +351,8 @@ public class JobExecutorService {
         }
     }
 
-    /**
-     * Handle different result types for synchronous jobs
-     *
-     * @param result The result object
-     * @return The appropriate ResponseEntity
-     * @throws IOException If there is an error processing the result
-     */
     private ResponseEntity<?> handleResultForSyncJob(Object result) throws IOException {
         if (result instanceof byte[]) {
-            // Return byte array as PDF
             return ResponseEntity.ok()
                     .contentType(MediaType.APPLICATION_PDF)
                     .header(
@@ -463,7 +360,6 @@ public class JobExecutorService {
                             "form-data; name=\"attachment\"; filename=\"result.pdf\"")
                     .body(result);
         } else if (result instanceof MultipartFile file) {
-            // Return MultipartFile content
             return ResponseEntity.ok()
                     .contentType(MediaType.parseMediaType(file.getContentType()))
                     .header(
@@ -473,7 +369,6 @@ public class JobExecutorService {
                                     + "\"")
                     .body(file.getBytes());
         } else {
-            // Default case: return as JSON
             return ResponseEntity.ok(result);
         }
     }
@@ -493,15 +388,9 @@ public class JobExecutorService {
         return mediaType != null ? mediaType.toString() : MediaType.APPLICATION_PDF_VALUE;
     }
 
-    /**
-     * Parse session timeout string (e.g., "30m", "1h") to milliseconds
-     *
-     * @param timeout The timeout string
-     * @return The timeout in milliseconds
-     */
     private long parseSessionTimeout(String timeout) {
         if (timeout == null || timeout.isEmpty()) {
-            return 30 * 60 * 1000; // Default: 30 minutes
+            return 30 * 60 * 1000;
         }
 
         try {
@@ -523,27 +412,16 @@ public class JobExecutorService {
                 case "m" -> (long) (numericValue * 60 * 1000);
                 case "h" -> (long) (numericValue * 60 * 60 * 1000);
                 case "d" -> (long) (numericValue * 24 * 60 * 60 * 1000);
-                default -> (long) (numericValue * 60 * 1000); // Default to minutes
+                default -> (long) (numericValue * 60 * 1000);
             };
         } catch (Exception e) {
             log.warn("Could not parse session timeout '{}', using default", timeout);
-            return 30 * 60 * 1000; // Default: 30 minutes
+            return 30 * 60 * 1000;
         }
     }
 
-    /**
-     * Execute a supplier with a timeout
-     *
-     * @param supplier The supplier to execute
-     * @param timeoutMs The timeout in milliseconds
-     * @return The result from the supplier
-     * @throws TimeoutException If the execution times out
-     * @throws Exception If the supplier throws an exception
-     */
     private <T> T executeWithTimeout(Supplier<T> supplier, long timeoutMs)
             throws TimeoutException, Exception {
-        // Use the same executor as other async jobs for consistency
-        // This ensures all operations run on the same thread pool
         String currentJobId = stirling.software.common.util.JobContext.getJobId();
 
         java.util.concurrent.CompletableFuture<T> future =
@@ -577,17 +455,10 @@ public class JobExecutorService {
         }
     }
 
-    /**
-     * Get a scoped job key that includes user ownership when security is enabled.
-     *
-     * @param baseJobId the base job identifier
-     * @return scoped job key, or just baseJobId if no ownership service available
-     */
     private String getScopedJobKey(String baseJobId) {
         if (jobOwnershipService != null) {
             return jobOwnershipService.createScopedJobKey(baseJobId);
         }
-        // Security disabled, return unsecured job key
         return baseJobId;
     }
 }

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

@@ -3,9 +3,13 @@ package stirling.software.common.service;
 import java.io.BufferedInputStream;
 import java.io.IOException;
 import java.io.InputStream;
+import java.time.Duration;
+import java.time.Instant;
 import java.time.LocalDateTime;
+import java.time.ZoneId;
 import java.time.temporal.ChronoUnit;
 import java.util.ArrayList;
+import java.util.HashMap;
 import java.util.List;
 import java.util.Locale;
 import java.util.Map;
@@ -16,6 +20,7 @@ import java.util.concurrent.TimeUnit;
 import java.util.zip.ZipEntry;
 import java.util.zip.ZipInputStream;
 
+import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.beans.factory.annotation.Value;
 import org.springframework.http.MediaType;
 import org.springframework.stereotype.Service;
@@ -26,6 +31,10 @@ import jakarta.annotation.PreDestroy;
 
 import lombok.extern.slf4j.Slf4j;
 
+import stirling.software.common.cluster.ClusterBackplane;
+import stirling.software.common.cluster.JobStore;
+import stirling.software.common.cluster.JobStoreEntry;
+import stirling.software.common.cluster.JobStoreEntry.JobState;
 import stirling.software.common.model.job.JobResult;
 import stirling.software.common.model.job.JobStats;
 import stirling.software.common.model.job.ResultFile;
@@ -40,20 +49,20 @@ public class TaskManager {
     private int jobResultExpiryMinutes = 30;
 
     private final FileStorage fileStorage;
+    private final JobStore jobStore;
+    private final ClusterBackplane clusterBackplane;
     private final ScheduledExecutorService cleanupExecutor =
             Executors.newSingleThreadScheduledExecutor(
                     Thread.ofVirtual().name("task-cleanup-", 0).factory());
 
-    /** Initialize the task manager and start the cleanup scheduler */
-    public TaskManager(FileStorage fileStorage) {
+    @Autowired
+    public TaskManager(
+            FileStorage fileStorage, JobStore jobStore, ClusterBackplane clusterBackplane) {
         this.fileStorage = fileStorage;
+        this.jobStore = jobStore;
+        this.clusterBackplane = clusterBackplane;
 
-        // Schedule periodic cleanup of old job results
-        cleanupExecutor.scheduleAtFixedRate(
-                this::cleanupOldJobs,
-                10, // Initial delay
-                10, // Interval
-                TimeUnit.MINUTES);
+        cleanupExecutor.scheduleAtFixedRate(this::cleanupOldJobs, 10, 10, TimeUnit.MINUTES);
 
         log.debug(
                 "Task manager initialized with job result expiry of {} minutes",
@@ -66,7 +75,9 @@ public class TaskManager {
      * @param jobId The job ID
      */
     public void createTask(String jobId) {
-        jobResults.put(jobId, JobResult.createNew(jobId));
+        JobResult result = JobResult.createNew(jobId);
+        jobResults.put(jobId, result);
+        writeThrough(jobId, result);
         log.debug("Created task with job ID: {}", jobId);
     }
 
@@ -79,6 +90,7 @@ public class TaskManager {
     public void setResult(String jobId, Object result) {
         JobResult jobResult = getOrCreateJobResult(jobId);
         jobResult.completeWithResult(result);
+        writeThrough(jobId, jobResult);
         log.debug("Set result for job ID: {}", jobId);
     }
 
@@ -101,6 +113,7 @@ public class TaskManager {
                         extractZipToIndividualFiles(fileId, originalFileName);
                 if (!extractedFiles.isEmpty()) {
                     jobResult.completeWithFiles(extractedFiles);
+                    writeThrough(jobId, jobResult);
                     log.debug(
                             "Set multiple file results for job ID: {} with {} files extracted from"
                                     + " ZIP",
@@ -127,6 +140,7 @@ public class TaskManager {
                     "Failed to get file size for job {}: {}. Using size 0.", jobId, e.getMessage());
             jobResult.completeWithSingleFile(fileId, originalFileName, contentType, 0);
         }
+        writeThrough(jobId, jobResult);
     }
 
     /**
@@ -138,6 +152,7 @@ public class TaskManager {
     public void setMultipleFileResults(String jobId, List<ResultFile> resultFiles) {
         JobResult jobResult = getOrCreateJobResult(jobId);
         jobResult.completeWithFiles(resultFiles);
+        writeThrough(jobId, jobResult);
         log.debug(
                 "Set multiple file results for job ID: {} with {} files",
                 jobId,
@@ -153,6 +168,7 @@ public class TaskManager {
     public void setError(String jobId, String error) {
         JobResult jobResult = getOrCreateJobResult(jobId);
         jobResult.failWithError(error);
+        writeThrough(jobId, jobResult);
         log.debug("Set error for job ID: {}: {}", jobId, error);
     }
 
@@ -169,6 +185,7 @@ public class TaskManager {
             // If no result or error has been set, mark it as complete with an empty result
             jobResult.completeWithResult("Task completed successfully");
         }
+        writeThrough(jobId, jobResult);
         log.debug("Marked job ID: {} as complete", jobId);
     }
 
@@ -205,6 +222,7 @@ public class TaskManager {
         JobResult jobResult = jobResults.get(jobId);
         if (jobResult != null) {
             jobResult.addNote(note);
+            writeThrough(jobId, jobResult);
             log.debug("Added note to job ID: {}: {}", jobId, note);
             return true;
         }
@@ -295,8 +313,11 @@ public class TaskManager {
         return jobResults.computeIfAbsent(jobId, JobResult::createNew);
     }
 
-    /** Clean up old completed job results */
+    /** Clean up old completed job results. No-op in cluster mode; the backplane TTL owns expiry. */
     public void cleanupOldJobs() {
+        if (clusterBackplane != null && !clusterBackplane.shouldRunLocalCleanup()) {
+            return;
+        }
         LocalDateTime expiryThreshold =
                 LocalDateTime.now().minus(jobResultExpiryMinutes, ChronoUnit.MINUTES);
         int removedCount = 0;
@@ -315,6 +336,9 @@ public class TaskManager {
 
                     // Remove the job result
                     jobResults.remove(entry.getKey());
+                    if (jobStore != null) {
+                        jobStore.delete(entry.getKey());
+                    }
                     removedCount++;
                 }
             }
@@ -327,6 +351,53 @@ public class TaskManager {
         }
     }
 
+    /** Mirror the in-memory {@code JobResult} into the cluster-visible {@link JobStore}. */
+    private void writeThrough(String jobId, JobResult result) {
+        if (jobStore == null) {
+            return;
+        }
+        try {
+            jobStore.put(toEntry(jobId, result), Duration.ofMinutes(jobResultExpiryMinutes));
+        } catch (RuntimeException ex) {
+            log.warn("JobStore write-through failed for job {}: {}", jobId, ex.getMessage());
+        }
+    }
+
+    private JobStoreEntry toEntry(String jobId, JobResult result) {
+        JobState state;
+        if (result.isComplete()) {
+            state = result.getError() != null ? JobState.FAILED : JobState.COMPLETE;
+        } else {
+            state = JobState.PENDING;
+        }
+        Instant createdAt = toInstant(result.getCreatedAt());
+        Instant completedAt = toInstant(result.getCompletedAt());
+        List<String> fileIds = new ArrayList<>();
+        if (result.hasFiles()) {
+            for (ResultFile rf : result.getAllResultFiles()) {
+                fileIds.add(rf.getFileId());
+            }
+        }
+        Map<String, String> meta = new HashMap<>();
+        if (result.getNotes() != null && !result.getNotes().isEmpty()) {
+            meta.put("notesCount", Integer.toString(result.getNotes().size()));
+        }
+        String owningNodeId = clusterBackplane == null ? "local" : clusterBackplane.localNodeId();
+        return new JobStoreEntry(
+                jobId,
+                state,
+                owningNodeId,
+                createdAt,
+                completedAt,
+                result.getError(),
+                fileIds,
+                meta);
+    }
+
+    private Instant toInstant(LocalDateTime ldt) {
+        return ldt == null ? null : ldt.atZone(ZoneId.systemDefault()).toInstant();
+    }
+
     /** Shutdown the cleanup executor */
     @PreDestroy
     public void shutdown() {
@@ -370,7 +441,7 @@ public class TaskManager {
             while ((entry = zipIn.getNextEntry()) != null) {
                 if (!entry.isDirectory()) {
                     String contentType = determineContentType(entry.getName());
-                    // storeInputStream returns the fileId and byte count — no extra stat needed
+                    // storeInputStream returns the fileId and byte count - no extra stat needed
                     FileStorage.StoredFile stored =
                             fileStorage.storeInputStream(zipIn, entry.getName());
 
@@ -458,7 +529,8 @@ public class TaskManager {
     }
 
     /**
-     * Find the job key that owns a given file ID.
+     * Find the job key that owns a given file ID. Checks the local in-memory map first, then falls
+     * back to the cluster-visible {@link JobStore}.
      *
      * @param fileId file identifier to look up
      * @return scoped job key if found, otherwise null
@@ -474,6 +546,18 @@ public class TaskManager {
                 }
             }
         }
+        if (jobStore != null) {
+            // Propagate JobStore failures: returning null on a backplane outage would conflate
+            // "no such file" with "lookup unavailable" and the caller would respond 404 to a
+            // transient blip that should be retried. Let Spring's exception handler surface a
+            // 5xx so clients know to retry.
+            try {
+                return jobStore.findJobIdByFileId(fileId).orElse(null);
+            } catch (RuntimeException e) {
+                log.warn("JobStore findJobIdByFileId failed for {}: {}", fileId, e.getMessage());
+                throw e;
+            }
+        }
         return null;
     }
 }

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

@@ -83,7 +83,16 @@ public class RequestUriUtils {
             return false;
         }
 
-        // Blocklist of backend/non-frontend paths that should still go through filters
+        // Blocklist of backend/non-frontend paths that should still go through filters.
+        //
+        // `/files` was historically a backend route; it is now a frontend route
+        // owned by HomePage / FileManagerView. Direct-nav or refresh on /files
+        // (or /files/<folder-uuid>) was returning the Spring auth filter's 401
+        // JSON instead of serving index.html, so the SPA never got a chance to
+        // mount and the user saw a raw error response. There are no `/files`
+        // backend mappings at the servlet root - the real storage endpoints
+        // live under `/api/v1/storage/files`, which is filtered out a few lines
+        // up by the `startsWith("/api/")` guard.
         String[] backendOnlyPrefixes = {
             "/register",
             "/pipeline",
@@ -91,7 +100,6 @@ public class RequestUriUtils {
             "/pdfjs-legacy",
             "/fonts",
             "/images",
-            "/files",
             "/css",
             "/js",
             "/swagger",
@@ -181,7 +189,7 @@ public class RequestUriUtils {
                 || trimmedUri.startsWith(
                         "/api/v1/mobile-scanner/") // Mobile scanner endpoints (no auth)
                 || trimmedUri.startsWith("/v1/api-docs")
-                // Workflow participant endpoints — access controlled by share tokens, not login
+                // Workflow participant endpoints - access controlled by share tokens, not login
                 || trimmedUri.startsWith("/api/v1/workflow/participant/")
                 // Share-link SPA bootstrap; data APIs remain protected
                 || trimmedUri.matches("^/share/[^/]+/?$");

app/common/src/main/java/stirling/software/common/util/propertyeditor/JsonListPropertyEditor.java

@@ -0,0 +1,48 @@
+package stirling.software.common.util.propertyeditor;
+
+import java.beans.PropertyEditorSupport;
+import java.util.ArrayList;
+import java.util.List;
+
+import lombok.extern.slf4j.Slf4j;
+
+import tools.jackson.core.type.TypeReference;
+import tools.jackson.databind.DeserializationFeature;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.json.JsonMapper;
+
+/**
+ * Binds a multipart form value containing a JSON array into a typed {@code List<T>}. Used for
+ * endpoints that accept structured list parameters via {@code @ModelAttribute} — the form field
+ * carries the full JSON array as its value and the editor parses it once.
+ */
+@Slf4j
+public class JsonListPropertyEditor<T> extends PropertyEditorSupport {
+
+    private static final ObjectMapper OBJECT_MAPPER =
+            JsonMapper.builder()
+                    .enable(DeserializationFeature.ACCEPT_SINGLE_VALUE_AS_ARRAY)
+                    .disable(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES)
+                    .build();
+
+    private final TypeReference<? extends List<T>> typeRef;
+
+    public JsonListPropertyEditor(TypeReference<? extends List<T>> typeRef) {
+        this.typeRef = typeRef;
+    }
+
+    @Override
+    public void setAsText(String text) throws IllegalArgumentException {
+        if (text == null || text.trim().isEmpty()) {
+            setValue(new ArrayList<T>());
+            return;
+        }
+        try {
+            setValue(OBJECT_MAPPER.readValue(text, typeRef));
+        } catch (Exception e) {
+            log.error("Failed to parse JSON list value", e);
+            throw new IllegalArgumentException(
+                    "Expected a JSON array but could not parse: " + e.getMessage());
+        }
+    }
+}

app/common/src/main/java/stirling/software/common/util/propertyeditor/JsonObjectPropertyEditor.java

@@ -0,0 +1,41 @@
+package stirling.software.common.util.propertyeditor;
+
+import java.beans.PropertyEditorSupport;
+
+import lombok.extern.slf4j.Slf4j;
+
+import tools.jackson.databind.DeserializationFeature;
+import tools.jackson.databind.ObjectMapper;
+import tools.jackson.databind.json.JsonMapper;
+
+/**
+ * Binds a multipart form value containing a JSON object into a typed {@code T}. Companion to {@link
+ * JsonListPropertyEditor} for single-object nested fields on {@code @ModelAttribute} endpoints.
+ */
+@Slf4j
+public class JsonObjectPropertyEditor<T> extends PropertyEditorSupport {
+
+    private static final ObjectMapper OBJECT_MAPPER =
+            JsonMapper.builder().disable(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES).build();
+
+    private final Class<T> type;
+
+    public JsonObjectPropertyEditor(Class<T> type) {
+        this.type = type;
+    }
+
+    @Override
+    public void setAsText(String text) throws IllegalArgumentException {
+        if (text == null || text.trim().isEmpty()) {
+            setValue(null);
+            return;
+        }
+        try {
+            setValue(OBJECT_MAPPER.readValue(text, type));
+        } catch (Exception e) {
+            log.error("Failed to parse JSON object value", e);
+            throw new IllegalArgumentException(
+                    "Expected a JSON object but could not parse: " + e.getMessage());
+        }
+    }
+}

app/common/src/main/java/stirling/software/common/util/propertyeditor/StringToArrayListPropertyEditor.java

@@ -1,53 +0,0 @@
-package stirling.software.common.util.propertyeditor;
-
-import java.beans.PropertyEditorSupport;
-import java.util.ArrayList;
-import java.util.List;
-
-import lombok.extern.slf4j.Slf4j;
-
-import tools.jackson.databind.DeserializationFeature;
-import tools.jackson.databind.JavaType;
-import tools.jackson.databind.ObjectMapper;
-import tools.jackson.databind.json.JsonMapper;
-
-/**
- * Spring property editor that decodes a JSON string into a typed {@link ArrayList}. Used to bind
- * complex list parameters (e.g. {@code List<RedactionArea>}, {@code List<EditTextOperation>}) from
- * multipart form fields, where Spring's default binding cannot deserialize a JSON array.
- */
-@Slf4j
-public class StringToArrayListPropertyEditor<T> extends PropertyEditorSupport {
-
-    private final ObjectMapper objectMapper =
-            JsonMapper.builder()
-                    .enable(DeserializationFeature.ACCEPT_SINGLE_VALUE_AS_ARRAY)
-                    .disable(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES)
-                    .build();
-
-    private final Class<T> elementType;
-
-    public StringToArrayListPropertyEditor(Class<T> elementType) {
-        this.elementType = elementType;
-    }
-
-    @Override
-    public void setAsText(String text) throws IllegalArgumentException {
-        if (text == null || text.trim().isEmpty()) {
-            setValue(new ArrayList<>());
-            return;
-        }
-        try {
-            JavaType listType =
-                    objectMapper
-                            .getTypeFactory()
-                            .constructCollectionType(ArrayList.class, elementType);
-            List<T> list = objectMapper.readValue(text, listType);
-            setValue(list);
-        } catch (Exception e) {
-            log.error("Exception while converting {}", e);
-            throw new IllegalArgumentException(
-                    "Failed to convert java.lang.String to java.util.List");
-        }
-    }
-}

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

@@ -0,0 +1,149 @@
+package stirling.software.SPDF.pdf.parser;
+
+import static org.assertj.core.api.Assertions.assertThat;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import org.junit.jupiter.api.Test;
+
+/** Unit tests for {@link PageColumnLayout} gutter detection and column classification. */
+class PageColumnLayoutTest {
+
+    private static final float PAGE_WIDTH = 612f; // Letter portrait
+
+    // ── single-column ────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void singleColumn_oneColumnNoGutters() {
+        List<float[]> lines = List.of(lineBox(72f, 396f));
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+
+        assertThat(layout.columnCount()).isEqualTo(1);
+        assertThat(layout.gutters()).isEmpty();
+    }
+
+    @Test
+    void singleColumn_classifyAnchor_returnsZero() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(List.of(lineBox(72f, 396f)), PAGE_WIDTH);
+
+        assertThat(layout.columnOf(100f, 200f)).isEqualTo(0);
+    }
+
+    // ── two-column ───────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void twoColumn_detectsGutter() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(buildTwoColumnLines(3), PAGE_WIDTH);
+
+        assertThat(layout.columnCount()).isEqualTo(2);
+        assertThat(layout.gutters()).hasSize(1);
+        float[] gutter = layout.gutters().get(0);
+        // Gutter is centered on pageWidth/2 with PageColumnLayout.MIDPOINT_SLACK_PT slack each
+        // side.
+        float pageMid = PAGE_WIDTH / 2f;
+        assertThat(gutter[0]).isBetween(pageMid - 40f, pageMid - 20f);
+        assertThat(gutter[1]).isBetween(pageMid + 20f, pageMid + 40f);
+    }
+
+    @Test
+    void twoColumn_classifyLeftAndRightAnchors() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(buildTwoColumnLines(3), PAGE_WIDTH);
+        assertThat(layout.columnOf(100f, 200f)).isEqualTo(0);
+        assertThat(layout.columnOf(380f, 460f)).isEqualTo(1);
+    }
+
+    @Test
+    void twoColumn_columnsCrossing_leftLineOnlyHitsLeft() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(buildTwoColumnLines(3), PAGE_WIDTH);
+        assertThat(layout.columnsCrossing(72f, 280f)).containsExactly(0);
+        assertThat(layout.columnsCrossing(320f, 540f)).containsExactly(1);
+    }
+
+    @Test
+    void twoColumn_spanningLine_returnsBothColumns() {
+        List<float[]> lines = new ArrayList<>(buildTwoColumnLines(3));
+        // Full-width header that crosses pageWidth/2.
+        lines.add(lineBox(72f, 396f));
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+
+        assertThat(layout.columnsCrossing(72f, 540f)).containsExactly(0, 1);
+        assertThat(layout.columnsCrossing(72f, 280f)).containsExactly(0);
+    }
+
+    private static List<float[]> buildTwoColumnLines(int rowsPerColumn) {
+        List<float[]> lines = new ArrayList<>();
+        for (int i = 0; i < rowsPerColumn; i++) {
+            lines.add(lineBox(72f, 136f)); // left column body (72..208)
+            lines.add(lineBox(320f, 220f)); // right column body (320..540)
+        }
+        return lines;
+    }
+
+    // ── three-column ─────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void threeColumn_collapsesToLeftRightSplit() {
+        // The midpoint-based detector splits the page at pageWidth/2 and treats anything else as
+        // single-column or spanning. A genuine 3-column layout collapses to 2 columns; the middle
+        // column's content ends up classified by midpoint as left or right of pageMid.
+        List<float[]> lines = new ArrayList<>();
+        for (int i = 0; i < 6; i++) {
+            lines.add(lineBox(72f, 150f)); // 72..222
+            lines.add(lineBox(252f, 150f)); // 252..402
+            lines.add(lineBox(432f, 150f)); // 432..582
+        }
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+
+        assertThat(layout.columnCount()).isEqualTo(2);
+        assertThat(layout.gutters()).hasSize(1);
+    }
+
+    // ── empty page ───────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void emptyPage_singleColumnFallback() {
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(List.of(), PAGE_WIDTH);
+        assertThat(layout.columnCount()).isEqualTo(1);
+        assertThat(layout.gutters()).isEmpty();
+    }
+
+    @Test
+    void onlyShortFragments_singleColumnFallback() {
+        // Page numbers / decorations — too narrow to vote either side.
+        List<float[]> lines = List.of(lineBox(300f, 6f));
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+        assertThat(layout.columnCount()).isEqualTo(1);
+    }
+
+    // ── narrow gap should not be confused for a gutter ───────────────────────────────────────────
+
+    @Test
+    void narrowInternalGap_doesNotProduceGutter() {
+        // Both halves sit left of the page midpoint, so no line votes for a right column and
+        // detection falls back to single-column.
+        List<float[]> lines = new ArrayList<>();
+        lines.add(lineBox(72f, 100f));
+        lines.add(lineBox(180f, 100f));
+        for (int i = 0; i < 5; i++) {
+            lines.add(lineBox(72f, 208f));
+        }
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+        assertThat(layout.columnCount()).isEqualTo(1);
+    }
+
+    // ── helpers ──────────────────────────────────────────────────────────────────────────────────
+
+    /** Builds a line bounding box {@code [x1, 0, x1+width, 0]}; Y is unused by detection. */
+    private static float[] lineBox(float x1, float width) {
+        return new float[] {x1, 0f, x1 + width, 0f};
+    }
+}

app/common/src/test/java/stirling/software/common/architecture/ArchitectureTest.java

@@ -56,4 +56,17 @@ class ArchitectureTest {
                         .resideInAPackage("stirling.software.saas..");
         rule.check(commonClasses);
     }
+
+    @Test
+    void clusterInterfacesHaveNoImplementationDependencies() {
+        ArchRule rule =
+                noClasses()
+                        .that()
+                        .resideInAPackage("stirling.software.common.cluster..")
+                        .should()
+                        .dependOnClassesThat()
+                        .resideInAnyPackage(
+                                "stirling.software.proprietary..", "stirling.software.saas..");
+        rule.check(commonClasses);
+    }
 }

app/common/src/test/java/stirling/software/common/cluster/BackplaneContractCompilationTest.java

@@ -0,0 +1,51 @@
+package stirling.software.common.cluster;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+import java.time.Instant;
+import java.util.List;
+import java.util.Map;
+
+import org.junit.jupiter.api.Test;
+
+class BackplaneContractCompilationTest {
+
+    @Test
+    void jobStoreEntryRecordRoundTrips() {
+        Instant now = Instant.now();
+        JobStoreEntry entry =
+                new JobStoreEntry(
+                        "job-1",
+                        JobStoreEntry.JobState.PENDING,
+                        "node-a",
+                        now,
+                        null,
+                        null,
+                        List.of("file-1"),
+                        Map.of("k", "v"));
+        assertEquals("job-1", entry.jobId());
+        assertEquals(JobStoreEntry.JobState.PENDING, entry.state());
+        assertEquals("node-a", entry.owningNodeId());
+        assertEquals(now, entry.createdAt());
+        assertEquals(List.of("file-1"), entry.fileIds());
+        assertEquals("v", entry.resultMeta().get("k"));
+    }
+
+    @Test
+    void clusterNodeRecordRoundTrips() {
+        Instant heartbeat = Instant.now();
+        ClusterNode node = new ClusterNode("node-a", "10.0.0.1:8080", heartbeat, "BOTH");
+        assertEquals("node-a", node.nodeId());
+        assertEquals("10.0.0.1:8080", node.internalAddress());
+        assertEquals(heartbeat, node.lastHeartbeat());
+        assertEquals("BOTH", node.role());
+    }
+
+    @Test
+    void rateLimitDecisionRecordRoundTrips() {
+        RateLimitStore.RateLimitDecision d = new RateLimitStore.RateLimitDecision(true, 7, 0L);
+        assertEquals(true, d.allowed());
+        assertEquals(7, d.remainingTokens());
+        assertEquals(0L, d.nanosToWaitForRefill());
+    }
+}

app/common/src/test/java/stirling/software/common/cluster/ClusterConfigValidationTest.java

@@ -0,0 +1,65 @@
+package stirling.software.common.cluster;
+
+import static org.junit.jupiter.api.Assertions.assertDoesNotThrow;
+import static org.junit.jupiter.api.Assertions.assertThrows;
+
+import java.lang.reflect.Method;
+
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.model.ApplicationProperties.Cluster;
+
+class ClusterConfigValidationTest {
+
+    @Test
+    void validationPassesWhenDisabled() {
+        ApplicationProperties props = new ApplicationProperties();
+        ClusterConfig config = new ClusterConfig(props);
+        assertDoesNotThrow(() -> invokeValidate(config));
+    }
+
+    @Test
+    void validationFailsWhenValkeyEnabledWithoutUrl() {
+        ApplicationProperties props = new ApplicationProperties();
+        Cluster cluster = props.getCluster();
+        cluster.setEnabled(true);
+        cluster.setBackplane("valkey");
+        ClusterConfig config = new ClusterConfig(props);
+        assertThrows(IllegalStateException.class, () -> invokeValidate(config));
+    }
+
+    @Test
+    void validationPassesWhenValkeyEnabledWithUrl() {
+        ApplicationProperties props = new ApplicationProperties();
+        Cluster cluster = props.getCluster();
+        cluster.setEnabled(true);
+        cluster.setBackplane("valkey");
+        cluster.getValkey().setUrl("redis://localhost:6379");
+        ClusterConfig config = new ClusterConfig(props);
+        assertDoesNotThrow(() -> invokeValidate(config));
+    }
+
+    @Test
+    void validationPassesWhenInProcessEnabled() {
+        ApplicationProperties props = new ApplicationProperties();
+        Cluster cluster = props.getCluster();
+        cluster.setEnabled(true);
+        cluster.setBackplane("inprocess");
+        ClusterConfig config = new ClusterConfig(props);
+        assertDoesNotThrow(() -> invokeValidate(config));
+    }
+
+    private void invokeValidate(ClusterConfig config) throws Exception {
+        Method m = ClusterConfig.class.getDeclaredMethod("validate");
+        m.setAccessible(true);
+        try {
+            m.invoke(config);
+        } catch (java.lang.reflect.InvocationTargetException ex) {
+            if (ex.getCause() instanceof RuntimeException re) {
+                throw re;
+            }
+            throw ex;
+        }
+    }
+}

app/common/src/test/java/stirling/software/common/cluster/ClusterPropertiesTest.java

@@ -0,0 +1,62 @@
+package stirling.software.common.cluster;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertFalse;
+import static org.junit.jupiter.api.Assertions.assertNotNull;
+
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.model.ApplicationProperties.Cluster;
+
+class ClusterPropertiesTest {
+
+    @Test
+    void defaultsAreDisabledAndInprocess() {
+        Cluster props = new ApplicationProperties().getCluster();
+        assertFalse(props.isEnabled());
+        assertEquals("inprocess", props.getBackplane());
+        assertEquals("local", props.getArtifactStore());
+        assertEquals(Cluster.NodeRole.BOTH, props.resolvedRole());
+        assertEquals("", props.getValkey().getUrl());
+        assertFalse(props.getValkey().getTls().isSkipCertVerification());
+        assertEquals("both", props.getNode().getRole());
+        assertEquals("http", props.getNode().getScheme());
+        assertEquals(5000L, props.getNode().getHeartbeatIntervalMs());
+    }
+
+    @Test
+    void resolvedRoleParsesCaseInsensitively() {
+        Cluster props = new ApplicationProperties().getCluster();
+        props.getNode().setRole("WEB");
+        assertEquals(Cluster.NodeRole.WEB, props.resolvedRole());
+
+        props.getNode().setRole("web");
+        assertEquals(Cluster.NodeRole.WEB, props.resolvedRole());
+
+        props.getNode().setRole("Worker");
+        assertEquals(Cluster.NodeRole.WORKER, props.resolvedRole());
+
+        props.getNode().setRole("garbage");
+        assertEquals(Cluster.NodeRole.BOTH, props.resolvedRole());
+
+        props.getNode().setRole(null);
+        assertEquals(Cluster.NodeRole.BOTH, props.resolvedRole());
+    }
+
+    @Test
+    void resolvedNodeIdIsStableAcrossCalls() {
+        Cluster props = new ApplicationProperties().getCluster();
+        String first = props.resolvedNodeId();
+        String second = props.resolvedNodeId();
+        assertNotNull(first);
+        assertEquals(first, second);
+    }
+
+    @Test
+    void resolvedNodeIdHonoursExplicitId() {
+        Cluster props = new ApplicationProperties().getCluster();
+        props.getNode().setId("abc");
+        assertEquals("abc", props.resolvedNodeId());
+    }
+}

app/common/src/test/java/stirling/software/common/cluster/InProcessConfigurationConditionalTest.java

@@ -0,0 +1,90 @@
+package stirling.software.common.cluster;
+
+import static org.assertj.core.api.Assertions.assertThat;
+
+import org.junit.jupiter.api.Test;
+import org.springframework.boot.autoconfigure.context.PropertyPlaceholderAutoConfiguration;
+import org.springframework.boot.test.context.runner.ApplicationContextRunner;
+import org.springframework.context.annotation.Bean;
+import org.springframework.context.annotation.Configuration;
+
+import stirling.software.common.cluster.inprocess.InProcessClusterConfiguration;
+import stirling.software.common.model.ApplicationProperties;
+
+/**
+ * Verifies the {@link InProcessClusterConfiguration} conditional wiring: in-process beans wire when
+ * cluster mode is off or {@code backplane=inprocess}, and are skipped when {@code
+ * backplane=valkey}.
+ */
+class InProcessConfigurationConditionalTest {
+
+    private final ApplicationContextRunner runner =
+            new ApplicationContextRunner()
+                    .withConfiguration(
+                            org.springframework.boot.autoconfigure.AutoConfigurations.of(
+                                    PropertyPlaceholderAutoConfiguration.class))
+                    .withUserConfiguration(
+                            TestAppPropertiesConfig.class,
+                            ClusterConfig.class,
+                            InProcessClusterConfiguration.class);
+
+    @Test
+    void inProcessBeansWireWhenClusterDisabled() {
+        runner.run(
+                context ->
+                        assertThat(context)
+                                .hasNotFailed()
+                                .hasSingleBean(ClusterBackplane.class)
+                                .hasSingleBean(JobStore.class)
+                                .hasSingleBean(RateLimitStore.class)
+                                .hasSingleBean(DistributedLock.class)
+                                .hasSingleBean(KeyValueCache.class)
+                                .hasSingleBean(InstanceRegistry.class));
+    }
+
+    @Test
+    void inProcessBeansWireWhenEnabledWithInProcessBackplane() {
+        runner.withPropertyValues("cluster.enabled=true", "cluster.backplane=inprocess")
+                .run(
+                        context ->
+                                assertThat(context)
+                                        .hasNotFailed()
+                                        .hasSingleBean(ClusterBackplane.class)
+                                        .hasSingleBean(JobStore.class)
+                                        .hasSingleBean(RateLimitStore.class)
+                                        .hasSingleBean(DistributedLock.class)
+                                        .hasSingleBean(KeyValueCache.class)
+                                        .hasSingleBean(InstanceRegistry.class));
+    }
+
+    @Test
+    void inProcessBeansSkippedWhenEnabledWithDistributedBackplane() {
+        runner.withPropertyValues(
+                        "cluster.enabled=true",
+                        "cluster.backplane=valkey",
+                        "cluster.valkey.url=redis://localhost:6379")
+                .run(
+                        context ->
+                                assertThat(context)
+                                        .hasNotFailed()
+                                        .doesNotHaveBean(ClusterBackplane.class)
+                                        .doesNotHaveBean(JobStore.class)
+                                        .doesNotHaveBean(RateLimitStore.class)
+                                        .doesNotHaveBean(DistributedLock.class)
+                                        .doesNotHaveBean(KeyValueCache.class)
+                                        .doesNotHaveBean(InstanceRegistry.class));
+    }
+
+    /**
+     * Hand-rolled {@link ApplicationProperties} bean: the production class loads YAML at startup
+     * via a {@code @PostConstruct} hook that isn't appropriate for the slice runner, so we wire a
+     * defaults-only instance here.
+     */
+    @Configuration
+    static class TestAppPropertiesConfig {
+        @Bean
+        ApplicationProperties applicationProperties() {
+            return new ApplicationProperties();
+        }
+    }
+}

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

@@ -0,0 +1,169 @@
+package stirling.software.common.cluster.inprocess;
+
+import static org.junit.jupiter.api.Assertions.assertFalse;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+import java.time.Duration;
+import java.util.Optional;
+import java.util.concurrent.CountDownLatch;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.atomic.AtomicBoolean;
+
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.cluster.DistributedLock;
+
+class InProcessDistributedLockTest {
+
+    @Test
+    void acquireReleaseAcquire() {
+        DistributedLock lock = new InProcessDistributedLock();
+        DistributedLock.LockHandle h1 = lock.tryAcquire("k", Duration.ofSeconds(30)).orElseThrow();
+        h1.release();
+        assertTrue(lock.tryAcquire("k", Duration.ofSeconds(30)).isPresent());
+    }
+
+    @Test
+    void reentryFromSameThreadFails_parityWithValkey() {
+        // The Valkey impl refuses reentry (SET NX semantics); the in-process impl must match,
+        // otherwise code working in single-instance silently breaks in cluster mode.
+        DistributedLock lock = new InProcessDistributedLock();
+        DistributedLock.LockHandle h1 = lock.tryAcquire("k", Duration.ofSeconds(30)).orElseThrow();
+        Optional<DistributedLock.LockHandle> reentry = lock.tryAcquire("k", Duration.ofSeconds(30));
+        assertFalse(reentry.isPresent(), "in-process lock must be non-reentrant");
+        h1.release();
+        // After release, anyone can acquire again.
+        assertTrue(lock.tryAcquire("k", Duration.ofSeconds(30)).isPresent());
+    }
+
+    @Test
+    void secondAcquireFromAnotherThreadFails() throws InterruptedException {
+        DistributedLock lock = new InProcessDistributedLock();
+        DistributedLock.LockHandle h1 = lock.tryAcquire("k", Duration.ofSeconds(30)).orElseThrow();
+
+        CountDownLatch done = new CountDownLatch(1);
+        AtomicBoolean acquired = new AtomicBoolean(true);
+        Thread t =
+                new Thread(
+                        () -> {
+                            Optional<DistributedLock.LockHandle> attempt =
+                                    lock.tryAcquire("k", Duration.ofSeconds(30));
+                            acquired.set(attempt.isPresent());
+                            attempt.ifPresent(DistributedLock.LockHandle::release);
+                            done.countDown();
+                        });
+        t.start();
+        assertTrue(done.await(2, TimeUnit.SECONDS));
+        assertFalse(acquired.get());
+        h1.release();
+    }
+
+    @Test
+    void leaseExpiryAllowsTakeoverEvenWithoutRelease() throws InterruptedException {
+        // Acquire with a short lease, never call release, then try to acquire again after the
+        // lease has elapsed. Matches Redis SET-NX-EX semantics - the second caller gets the lock
+        // because the first lease auto-expired. 250ms lease + 350ms wait gives CI generous slack.
+        DistributedLock lock = new InProcessDistributedLock();
+        DistributedLock.LockHandle h1 = lock.tryAcquire("k", Duration.ofMillis(250)).orElseThrow();
+        Thread.sleep(350);
+        Optional<DistributedLock.LockHandle> takeover =
+                lock.tryAcquire("k", Duration.ofSeconds(30));
+        assertTrue(
+                takeover.isPresent(),
+                "expired lease must release the lock so a new caller can take over");
+        // Calling release() on the original handle after takeover must be a no-op (token check).
+        h1.release();
+        // The takeover holder is still the legitimate owner.
+        assertFalse(lock.tryAcquire("k", Duration.ofSeconds(30)).isPresent());
+        takeover.get().release();
+    }
+
+    @Test
+    void renewExtendsLease() throws InterruptedException {
+        // Acquire with a short lease, renew it before it expires, then verify the lock is still
+        // held past the original expiry point. 200ms initial + renew to 2s + wait 350ms.
+        DistributedLock lock = new InProcessDistributedLock();
+        DistributedLock.LockHandle h1 = lock.tryAcquire("k", Duration.ofMillis(200)).orElseThrow();
+        assertTrue(h1.renew(Duration.ofSeconds(2)), "renew on a held lease must succeed");
+        Thread.sleep(350);
+        assertFalse(
+                lock.tryAcquire("k", Duration.ofSeconds(30)).isPresent(),
+                "renew should have pushed expiry well past the original 200ms");
+        h1.release();
+    }
+
+    @Test
+    void renewAfterReleaseFails() {
+        DistributedLock lock = new InProcessDistributedLock();
+        DistributedLock.LockHandle h1 = lock.tryAcquire("k", Duration.ofSeconds(30)).orElseThrow();
+        h1.release();
+        assertFalse(h1.renew(Duration.ofSeconds(30)), "renew on a released handle must fail");
+    }
+
+    /**
+     * Concurrency stress: many threads contending on the same key with each holder respecting the
+     * lease (hold &lt;&lt; lease). The lock behaves as a strict mutex in this regime so asserting
+     * mutual exclusion is meaningful. A separate test ({@link
+     * #leaseExpiryAllowsTakeoverEvenWithoutRelease}) covers the takeover-across-expiry branch,
+     * which legitimately allows two holders momentarily and is split-brain behaviour inherent to
+     * any lease-based lock.
+     */
+    @Test
+    void concurrentContentionPreservesMutualExclusion() throws InterruptedException {
+        DistributedLock lock = new InProcessDistributedLock();
+        int threads = 16;
+        int attemptsPerThread = 200;
+        // Lease far exceeds any plausible hold time, so the takeover branch never triggers in
+        // this test and the lock acts as a strict mutex.
+        Duration lease = Duration.ofSeconds(5);
+        java.util.concurrent.atomic.AtomicInteger concurrentHolders =
+                new java.util.concurrent.atomic.AtomicInteger();
+        java.util.concurrent.atomic.AtomicInteger maxConcurrent =
+                new java.util.concurrent.atomic.AtomicInteger();
+        java.util.concurrent.atomic.AtomicInteger acquires =
+                new java.util.concurrent.atomic.AtomicInteger();
+        java.util.concurrent.atomic.AtomicReference<Throwable> firstFailure =
+                new java.util.concurrent.atomic.AtomicReference<>();
+        CountDownLatch start = new CountDownLatch(1);
+        CountDownLatch done = new CountDownLatch(threads);
+
+        for (int i = 0; i < threads; i++) {
+            new Thread(
+                            () -> {
+                                try {
+                                    start.await();
+                                    for (int j = 0; j < attemptsPerThread; j++) {
+                                        Optional<DistributedLock.LockHandle> h =
+                                                lock.tryAcquire("hot", lease);
+                                        if (h.isPresent()) {
+                                            int now = concurrentHolders.incrementAndGet();
+                                            maxConcurrent.accumulateAndGet(now, Math::max);
+                                            acquires.incrementAndGet();
+                                            // Trivial critical section; well within lease.
+                                            concurrentHolders.decrementAndGet();
+                                            h.get().release();
+                                        }
+                                    }
+                                } catch (Throwable t) {
+                                    firstFailure.compareAndSet(null, t);
+                                } finally {
+                                    done.countDown();
+                                }
+                            },
+                            "lock-stress-" + i)
+                    .start();
+        }
+        start.countDown();
+        assertTrue(done.await(30, TimeUnit.SECONDS), "stress workers must finish in time");
+        org.junit.jupiter.api.Assertions.assertNull(firstFailure.get(), "no worker may throw");
+        org.junit.jupiter.api.Assertions.assertEquals(
+                1,
+                maxConcurrent.get(),
+                "mutual exclusion violated: more than one holder observed simultaneously");
+        assertTrue(
+                acquires.get() > 0,
+                "at least some acquires must succeed under contention (saw "
+                        + acquires.get()
+                        + ")");
+    }
+}

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

@@ -0,0 +1,28 @@
+package stirling.software.common.cluster.inprocess;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+import java.time.Duration;
+import java.time.Instant;
+
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.cluster.ClusterNode;
+
+class InProcessInstanceRegistryTest {
+
+    @Test
+    void registerThenLookupAndActiveNodes() {
+        InProcessInstanceRegistry registry = new InProcessInstanceRegistry();
+        ClusterNode node = new ClusterNode("node-1", "127.0.0.1:8080", Instant.now(), "BOTH");
+        registry.register(node, Duration.ofSeconds(30));
+
+        assertTrue(registry.lookup("node-1").isPresent());
+        assertEquals("node-1", registry.lookup("node-1").get().nodeId());
+        assertEquals(1, registry.activeNodes().size());
+
+        registry.deregister("node-1");
+        assertTrue(registry.lookup("node-1").isEmpty());
+    }
+}

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

@@ -0,0 +1,91 @@
+package stirling.software.common.cluster.inprocess;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertFalse;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+import java.time.Duration;
+import java.time.Instant;
+import java.util.List;
+import java.util.Map;
+
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.cluster.JobStoreEntry;
+
+class InProcessJobStoreTest {
+
+    private final InProcessJobStore store = new InProcessJobStore();
+
+    @Test
+    void putGetDeleteExistsRoundTrip() {
+        JobStoreEntry entry = entry("job-1");
+        store.put(entry, Duration.ofMinutes(30));
+
+        assertTrue(store.exists("job-1"));
+        assertEquals(entry, store.get("job-1").orElseThrow());
+
+        store.delete("job-1");
+        assertFalse(store.exists("job-1"));
+    }
+
+    @Test
+    void ttlExpiry() throws InterruptedException {
+        store.put(entry("job-2"), Duration.ofMillis(50));
+        Thread.sleep(100);
+        assertFalse(store.get("job-2").isPresent());
+    }
+
+    @Test
+    void purgeExpiredRemovesOnlyStaleEntries() throws InterruptedException {
+        store.put(entry("job-fresh"), Duration.ofMinutes(30));
+        store.put(entry("job-stale"), Duration.ofMillis(20));
+        Thread.sleep(80);
+
+        int removed = store.purgeExpired();
+        assertEquals(1, removed);
+        assertTrue(store.exists("job-fresh"));
+    }
+
+    @Test
+    void findJobIdByFileIdReturnsTheRightJob() {
+        store.put(
+                new JobStoreEntry(
+                        "job-a",
+                        JobStoreEntry.JobState.COMPLETE,
+                        "node-1",
+                        Instant.now(),
+                        Instant.now(),
+                        null,
+                        List.of("file-1", "file-2"),
+                        Map.of()),
+                Duration.ofMinutes(30));
+        store.put(
+                new JobStoreEntry(
+                        "job-b",
+                        JobStoreEntry.JobState.COMPLETE,
+                        "node-1",
+                        Instant.now(),
+                        Instant.now(),
+                        null,
+                        List.of("file-3"),
+                        Map.of()),
+                Duration.ofMinutes(30));
+
+        assertEquals("job-a", store.findJobIdByFileId("file-1").orElseThrow());
+        assertEquals("job-b", store.findJobIdByFileId("file-3").orElseThrow());
+        assertFalse(store.findJobIdByFileId("missing").isPresent());
+    }
+
+    private JobStoreEntry entry(String id) {
+        return new JobStoreEntry(
+                id,
+                JobStoreEntry.JobState.PENDING,
+                "node-1",
+                Instant.now(),
+                null,
+                null,
+                List.of(),
+                Map.of());
+    }
+}

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

@@ -0,0 +1,41 @@
+package stirling.software.common.cluster.inprocess;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertFalse;
+
+import java.time.Duration;
+
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.cluster.KeyValueCache;
+
+class InProcessKeyValueCacheTest {
+
+    @Test
+    void putGetEvict() {
+        KeyValueCache cache = new InProcessKeyValueCache();
+        cache.put("apikey", "a", "userA", Duration.ofMinutes(1));
+        assertEquals("userA", cache.get("apikey", "a").orElseThrow());
+
+        cache.evict("apikey", "a");
+        assertFalse(cache.get("apikey", "a").isPresent());
+    }
+
+    @Test
+    void ttlExpiry() throws InterruptedException {
+        KeyValueCache cache = new InProcessKeyValueCache();
+        cache.put("ns", "k", "v", Duration.ofMillis(40));
+        Thread.sleep(80);
+        assertFalse(cache.get("ns", "k").isPresent());
+    }
+
+    @Test
+    void evictNamespace() {
+        KeyValueCache cache = new InProcessKeyValueCache();
+        cache.put("ns", "a", "1", Duration.ofMinutes(1));
+        cache.put("ns", "b", "2", Duration.ofMinutes(1));
+        cache.evictNamespace("ns");
+        assertFalse(cache.get("ns", "a").isPresent());
+        assertFalse(cache.get("ns", "b").isPresent());
+    }
+}

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

@@ -0,0 +1,56 @@
+package stirling.software.common.cluster.inprocess;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertFalse;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+import java.time.Duration;
+
+import org.junit.jupiter.api.Test;
+
+import stirling.software.common.cluster.RateLimitStore;
+import stirling.software.common.cluster.RateLimitStore.RateLimitDecision;
+
+class InProcessRateLimitStoreTest {
+
+    @Test
+    void firstNConsumesAllowed() {
+        RateLimitStore store = new InProcessRateLimitStore();
+        for (int i = 0; i < 5; i++) {
+            assertTrue(store.tryConsume("k", 5, Duration.ofSeconds(60)).allowed(), "i=" + i);
+        }
+        assertFalse(store.tryConsume("k", 5, Duration.ofSeconds(60)).allowed());
+    }
+
+    @Test
+    void remainingTokensDecrements() {
+        RateLimitStore store = new InProcessRateLimitStore();
+        RateLimitDecision d1 = store.tryConsume("k", 5, Duration.ofSeconds(60));
+        RateLimitDecision d2 = store.tryConsume("k", 5, Duration.ofSeconds(60));
+        assertTrue(d1.allowed());
+        assertTrue(d2.allowed());
+        assertEquals(4, d1.remainingTokens());
+        assertEquals(3, d2.remainingTokens());
+    }
+
+    @Test
+    void refillRestoresTokens() throws InterruptedException {
+        RateLimitStore store = new InProcessRateLimitStore();
+        // Capacity 2 with smooth refill over 100 ms -> ~1 token per 50 ms.
+        for (int i = 0; i < 2; i++) {
+            assertTrue(store.tryConsume("k", 2, Duration.ofMillis(100)).allowed());
+        }
+        assertFalse(store.tryConsume("k", 2, Duration.ofMillis(100)).allowed());
+        Thread.sleep(150);
+        assertTrue(store.tryConsume("k", 2, Duration.ofMillis(100)).allowed());
+    }
+
+    @Test
+    void deniedConsumeReportsWaitNanos() {
+        RateLimitStore store = new InProcessRateLimitStore();
+        assertTrue(store.tryConsume("wait", 1, Duration.ofSeconds(10)).allowed());
+        RateLimitDecision denied = store.tryConsume("wait", 1, Duration.ofSeconds(10));
+        assertFalse(denied.allowed());
+        assertTrue(denied.nanosToWaitForRefill() > 0L);
+    }
+}

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

@@ -0,0 +1,42 @@
+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.assertThrows;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+import java.io.ByteArrayInputStream;
+import java.io.IOException;
+import java.nio.file.Path;
+
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.io.TempDir;
+
+import stirling.software.common.cluster.FileStore;
+
+class LocalDiskFileStoreTest {
+
+    @Test
+    void storeRetrieveSizeDeleteExistsRoundTrip(@TempDir Path dir) throws IOException {
+        LocalDiskFileStore store = new LocalDiskFileStore(dir.toString());
+        byte[] payload = "hello-bytes".getBytes();
+
+        FileStore.Stored stored = store.store(new ByteArrayInputStream(payload), "x.txt");
+        assertEquals(payload.length, stored.size());
+        assertTrue(store.exists(stored.fileId()));
+        assertEquals(payload.length, store.size(stored.fileId()));
+        assertArrayEquals(payload, store.retrieveBytes(stored.fileId()));
+
+        assertTrue(store.delete(stored.fileId()));
+        assertFalse(store.exists(stored.fileId()));
+    }
+
+    @Test
+    void traversalIdsAreRejected(@TempDir Path dir) {
+        LocalDiskFileStore store = new LocalDiskFileStore(dir.toString());
+        assertThrows(IllegalArgumentException.class, () -> store.resolve("../foo"));
+        assertThrows(IllegalArgumentException.class, () -> store.resolve("a/b"));
+        assertThrows(IllegalArgumentException.class, () -> store.resolve("a\\b"));
+    }
+}

app/common/src/test/java/stirling/software/common/configuration/ConfigInitializerTest.java

@@ -0,0 +1,95 @@
+package stirling.software.common.configuration;
+
+import static org.junit.jupiter.api.Assertions.*;
+
+import org.junit.jupiter.api.Test;
+import org.snakeyaml.engine.v2.api.LoadSettings;
+
+import stirling.software.common.util.YamlHelper;
+
+class ConfigInitializerTest {
+
+    private static final LoadSettings LOAD_SETTINGS =
+            LoadSettings.builder()
+                    .setUseMarks(true)
+                    .setMaxAliasesForCollections(Integer.MAX_VALUE)
+                    .setAllowRecursiveKeys(true)
+                    .setParseComments(true)
+                    .build();
+
+    // Mirrors the proFeatures block of settings.yml.template after the camelCase rename.
+    private static final String CAMEL_CASE_TEMPLATE =
+            """
+            premium:
+              proFeatures:
+                ssoAutoLogin: false
+                customMetadata:
+                  autoUpdateMetadata: false
+                  author: username
+                  creator: Stirling-PDF
+                  producer: Stirling-PDF
+            """;
+
+    @Test
+    void migrateProFeaturesKeyCasing_carriesForwardLegacyPascalCaseValues() {
+        // An existing install whose settings.yml still uses the old PascalCase keys.
+        String legacy =
+                """
+                premium:
+                  proFeatures:
+                    SSOAutoLogin: true
+                    CustomMetadata:
+                      autoUpdateMetadata: true
+                      author: alice
+                      creator: bob
+                      producer: carol
+                """;
+        YamlHelper template = new YamlHelper(LOAD_SETTINGS, CAMEL_CASE_TEMPLATE);
+        YamlHelper existing = new YamlHelper(LOAD_SETTINGS, legacy);
+
+        new ConfigInitializer().migrateProFeaturesKeyCasing(existing, template);
+
+        assertEquals(
+                "true", template.getValueByExactKeyPath("premium", "proFeatures", "ssoAutoLogin"));
+        assertEquals(
+                "true",
+                template.getValueByExactKeyPath(
+                        "premium", "proFeatures", "customMetadata", "autoUpdateMetadata"));
+        assertEquals(
+                "alice",
+                template.getValueByExactKeyPath(
+                        "premium", "proFeatures", "customMetadata", "author"));
+        assertEquals(
+                "bob",
+                template.getValueByExactKeyPath(
+                        "premium", "proFeatures", "customMetadata", "creator"));
+        assertEquals(
+                "carol",
+                template.getValueByExactKeyPath(
+                        "premium", "proFeatures", "customMetadata", "producer"));
+    }
+
+    @Test
+    void migrateProFeaturesKeyCasing_withoutLegacyKeys_keepsTemplateDefaults() {
+        // No PascalCase keys present -> this migration step must be a no-op.
+        String alreadyCamel =
+                """
+                premium:
+                  proFeatures:
+                    ssoAutoLogin: true
+                    customMetadata:
+                      author: dave
+                """;
+        YamlHelper template = new YamlHelper(LOAD_SETTINGS, CAMEL_CASE_TEMPLATE);
+        YamlHelper existing = new YamlHelper(LOAD_SETTINGS, alreadyCamel);
+
+        new ConfigInitializer().migrateProFeaturesKeyCasing(existing, template);
+
+        assertEquals(
+                "false", template.getValueByExactKeyPath("premium", "proFeatures", "ssoAutoLogin"));
+        assertEquals(
+                "username",
+                template.getValueByExactKeyPath(
+                        "premium", "proFeatures", "customMetadata", "author"));
+    }
+}

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

@@ -0,0 +1,27 @@
+package stirling.software.common.service;
+
+import static org.junit.jupiter.api.Assertions.assertArrayEquals;
+import static org.mockito.Mockito.mock;
+
+import java.io.IOException;
+import java.nio.file.Path;
+
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.io.TempDir;
+
+import stirling.software.common.cluster.inprocess.LocalDiskFileStore;
+
+class FileStorageDelegationTest {
+
+    @Test
+    void storeBytesThenRetrieveBytesRoundTripsThroughFileStore(@TempDir Path tempDir)
+            throws IOException {
+        FileStorage fs =
+                new FileStorage(
+                        mock(FileOrUploadService.class),
+                        new LocalDiskFileStore(tempDir.toString()));
+        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/FileStorageTest.java

@@ -3,6 +3,7 @@ package stirling.software.common.service;
 import static org.junit.jupiter.api.Assertions.*;
 import static org.mockito.Mockito.*;
 
+import java.io.ByteArrayInputStream;
 import java.io.IOException;
 import java.io.InputStream;
 import java.nio.charset.StandardCharsets;
@@ -13,29 +14,30 @@ import java.util.stream.Stream;
 import org.junit.jupiter.api.BeforeEach;
 import org.junit.jupiter.api.Test;
 import org.junit.jupiter.api.io.TempDir;
-import org.mockito.InjectMocks;
 import org.mockito.Mock;
 import org.mockito.MockitoAnnotations;
 import org.springframework.core.io.ByteArrayResource;
 import org.springframework.core.io.Resource;
 import org.springframework.http.MediaType;
-import org.springframework.test.util.ReflectionTestUtils;
 import org.springframework.web.multipart.MultipartFile;
 
+import stirling.software.common.cluster.inprocess.LocalDiskFileStore;
+
 class FileStorageTest {
 
     @TempDir Path tempDir;
 
     @Mock private FileOrUploadService fileOrUploadService;
 
-    @InjectMocks private FileStorage fileStorage;
+    private FileStorage fileStorage;
 
     private MultipartFile mockFile;
 
     @BeforeEach
-    void setUp() {
+    void setUp() throws IOException {
         MockitoAnnotations.openMocks(this);
-        ReflectionTestUtils.setField(fileStorage, "tempDirPath", tempDir.toString());
+        fileStorage =
+                new FileStorage(fileOrUploadService, new LocalDiskFileStore(tempDir.toString()));
 
         // Create a mock MultipartFile
         mockFile = mock(MultipartFile.class);
@@ -47,17 +49,7 @@ class FileStorageTest {
     void testStoreFile() throws IOException {
         // Arrange
         byte[] fileContent = "Test PDF content".getBytes();
-        when(mockFile.getBytes()).thenReturn(fileContent);
-
-        // Set up mock to handle transferTo by writing the file
-        doAnswer(
-                        invocation -> {
-                            java.io.File file = invocation.getArgument(0);
-                            Files.write(file.toPath(), fileContent);
-                            return null;
-                        })
-                .when(mockFile)
-                .transferTo(any(java.io.File.class));
+        when(mockFile.getInputStream()).thenReturn(new ByteArrayInputStream(fileContent));
 
         // Act
         String fileId = fileStorage.storeFile(mockFile);
@@ -65,7 +57,7 @@ class FileStorageTest {
         // Assert
         assertNotNull(fileId);
         assertTrue(Files.exists(tempDir.resolve(fileId)));
-        verify(mockFile).transferTo(any(java.io.File.class));
+        assertArrayEquals(fileContent, Files.readAllBytes(tempDir.resolve(fileId)));
     }
 
     @Test
@@ -247,11 +239,11 @@ class FileStorageTest {
             filesBefore = s.count();
         }
 
-        // Act + Assert: IOException must propagate out — not be swallowed.
+        // Act + Assert: IOException must propagate out - not be swallowed.
         assertThrows(
                 IOException.class, () -> fileStorage.storeFromResource(flakyResource, "n.pdf"));
 
-        // Assert: no partial file lingers under the storage directory — the finally
+        // Assert: no partial file lingers under the storage directory - the finally
         // branch's deleteIfExists must have cleaned it up.
         long filesAfter;
         try (Stream<Path> s = Files.list(tempDir)) {

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

@@ -0,0 +1,122 @@
+package stirling.software.common.service;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.mockito.ArgumentMatchers.any;
+import static org.mockito.Mockito.never;
+import static org.mockito.Mockito.spy;
+import static org.mockito.Mockito.verify;
+
+import java.time.LocalDateTime;
+
+import org.junit.jupiter.api.BeforeEach;
+import org.junit.jupiter.api.Test;
+import org.mockito.Mock;
+import org.mockito.MockitoAnnotations;
+import org.springframework.test.util.ReflectionTestUtils;
+
+import stirling.software.common.cluster.ClusterBackplane;
+import stirling.software.common.cluster.JobStoreEntry;
+import stirling.software.common.cluster.JobStoreEntry.JobState;
+import stirling.software.common.cluster.inprocess.InProcessClusterBackplane;
+import stirling.software.common.cluster.inprocess.InProcessJobStore;
+import stirling.software.common.model.ApplicationProperties;
+import stirling.software.common.model.job.JobResult;
+
+class TaskManagerJobStoreDelegationTest {
+
+    @Mock private FileStorage fileStorage;
+
+    private InProcessJobStore jobStore;
+    private ClusterBackplane backplane;
+    private TaskManager taskManager;
+
+    @BeforeEach
+    void setUp() {
+        MockitoAnnotations.openMocks(this);
+        jobStore = spy(new InProcessJobStore());
+        backplane = new InProcessClusterBackplane(new ApplicationProperties());
+        taskManager = new TaskManager(fileStorage, jobStore, backplane);
+        ReflectionTestUtils.setField(taskManager, "jobResultExpiryMinutes", 30);
+    }
+
+    @Test
+    void createTaskWritesPendingEntry() {
+        taskManager.createTask("job-1");
+        JobStoreEntry entry = jobStore.get("job-1").orElseThrow();
+        assertEquals(JobState.PENDING, entry.state());
+        assertEquals(backplane.localNodeId(), entry.owningNodeId());
+    }
+
+    @Test
+    void setCompleteFlipsToComplete() {
+        taskManager.createTask("job-2");
+        taskManager.setResult("job-2", "ok");
+        taskManager.setComplete("job-2");
+        JobStoreEntry entry = jobStore.get("job-2").orElseThrow();
+        assertEquals(JobState.COMPLETE, entry.state());
+    }
+
+    @Test
+    void setErrorFlipsToFailed() {
+        taskManager.createTask("job-3");
+        taskManager.setError("job-3", "boom");
+        JobStoreEntry entry = jobStore.get("job-3").orElseThrow();
+        assertEquals(JobState.FAILED, entry.state());
+        assertEquals("boom", entry.error());
+    }
+
+    @Test
+    void cleanupOldJobsIsNoopWhenBackplaneIsNotInProcess() {
+        ClusterBackplane mockedValkeyBackplane =
+                new ClusterBackplane() {
+                    @Override
+                    public boolean isHealthy() {
+                        return true;
+                    }
+
+                    @Override
+                    public String backplaneType() {
+                        return "valkey";
+                    }
+
+                    @Override
+                    public String localNodeId() {
+                        return "node-1";
+                    }
+
+                    @Override
+                    public boolean shouldRunLocalCleanup() {
+                        // Distributed backplanes own job TTL eviction themselves; this mock
+                        // mirrors the real ValkeyClusterBackplane override of the default true.
+                        return false;
+                    }
+                };
+        TaskManager tm = new TaskManager(fileStorage, jobStore, mockedValkeyBackplane);
+        ReflectionTestUtils.setField(tm, "jobResultExpiryMinutes", 30);
+        tm.createTask("job-4");
+        tm.setComplete("job-4");
+        ageJobPastExpiry(tm, "job-4");
+        tm.cleanupOldJobs();
+        // cleanup must short-circuit before touching jobStore in cluster mode; the backplane
+        // TTL owns expiry there. If the gate fired correctly, delete is never called.
+        verify(jobStore, never()).delete(any());
+    }
+
+    @Test
+    void cleanupOldJobsDeletesFromJobStoreWhenBackplaneIsInProcess() {
+        taskManager.createTask("job-5");
+        taskManager.setComplete("job-5");
+        ageJobPastExpiry(taskManager, "job-5");
+        taskManager.cleanupOldJobs();
+        verify(jobStore).delete("job-5");
+    }
+
+    @SuppressWarnings("unchecked")
+    private static void ageJobPastExpiry(TaskManager tm, String jobId) {
+        var jobResults =
+                (java.util.Map<String, JobResult>) ReflectionTestUtils.getField(tm, "jobResults");
+        JobResult result = jobResults.get(jobId);
+        ReflectionTestUtils.setField(result, "completedAt", LocalDateTime.now().minusHours(2));
+        ReflectionTestUtils.setField(result, "complete", true);
+    }
+}

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

@@ -1,20 +1,28 @@
 package stirling.software.common.service;
 
 import static org.junit.jupiter.api.Assertions.*;
+import static org.mockito.ArgumentMatchers.any;
+import static org.mockito.ArgumentMatchers.anyString;
 import static org.mockito.Mockito.*;
 
 import java.time.LocalDateTime;
 import java.util.Map;
+import java.util.Optional;
 
 import org.junit.jupiter.api.AfterEach;
 import org.junit.jupiter.api.BeforeEach;
 import org.junit.jupiter.api.Test;
+import org.mockito.ArgumentCaptor;
 import org.mockito.InjectMocks;
 import org.mockito.Mock;
 import org.mockito.MockitoAnnotations;
 import org.springframework.http.MediaType;
 import org.springframework.test.util.ReflectionTestUtils;
 
+import stirling.software.common.cluster.ClusterBackplane;
+import stirling.software.common.cluster.JobStore;
+import stirling.software.common.cluster.JobStoreEntry;
+import stirling.software.common.cluster.JobStoreEntry.JobState;
 import stirling.software.common.model.job.JobResult;
 import stirling.software.common.model.job.JobStats;
 import stirling.software.common.model.job.ResultFile;
@@ -22,6 +30,8 @@ import stirling.software.common.model.job.ResultFile;
 class TaskManagerTest {
 
     @Mock private FileStorage fileStorage;
+    @Mock private JobStore jobStore;
+    @Mock private ClusterBackplane clusterBackplane;
 
     @InjectMocks private TaskManager taskManager;
 
@@ -30,6 +40,10 @@ class TaskManagerTest {
     @BeforeEach
     void setUp() {
         closeable = MockitoAnnotations.openMocks(this);
+        // Treat the backplane as in-process so cleanupOldJobs is not short-circuited.
+        lenient().when(clusterBackplane.backplaneType()).thenReturn("inprocess");
+        lenient().when(clusterBackplane.localNodeId()).thenReturn("test-node");
+        lenient().when(clusterBackplane.shouldRunLocalCleanup()).thenReturn(true);
         ReflectionTestUtils.setField(taskManager, "jobResultExpiryMinutes", 30);
     }
 
@@ -270,6 +284,33 @@ class TaskManagerTest {
         verify(fileStorage).deleteFile("file-id");
     }
 
+    @Test
+    void testCleanupOldJobs_NoOpWhenBackplaneOwnsExpiry() {
+        // When the backplane reports it should NOT run local cleanup (e.g. a distributed
+        // backplane with its own TTL), the cleanup loop must leave local state untouched.
+        when(clusterBackplane.shouldRunLocalCleanup()).thenReturn(false);
+
+        // Seed an old completed job that would normally be removed.
+        String oldJobId = "old-job-distributed";
+        taskManager.createTask(oldJobId);
+        JobResult oldJob = taskManager.getJobResult(oldJobId);
+        ReflectionTestUtils.setField(oldJob, "completedAt", LocalDateTime.now().minusHours(1));
+        ReflectionTestUtils.setField(oldJob, "complete", true);
+
+        Map<String, JobResult> jobResultsMap =
+                (Map<String, JobResult>) ReflectionTestUtils.getField(taskManager, "jobResults");
+        assertNotNull(jobResultsMap);
+        assertTrue(jobResultsMap.containsKey(oldJobId));
+
+        // Act
+        taskManager.cleanupOldJobs();
+
+        // Assert: nothing was removed locally, and no jobStore.delete was issued.
+        assertTrue(jobResultsMap.containsKey(oldJobId));
+        verify(jobStore, never()).delete(anyString());
+        verify(fileStorage, never()).deleteFile(anyString());
+    }
+
     @Test
     void testShutdown() {
         // This mainly tests that the shutdown method doesn't throw exceptions
@@ -310,4 +351,33 @@ class TaskManagerTest {
         // Assert
         assertFalse(result);
     }
+
+    @Test
+    void testWriteThroughOnUpdate() {
+        // Mutating calls must write through to the injected JobStore.
+        String jobId = "write-through-job";
+        taskManager.createTask(jobId);
+        taskManager.setResult(jobId, "done");
+
+        ArgumentCaptor<JobStoreEntry> captor = ArgumentCaptor.forClass(JobStoreEntry.class);
+        verify(jobStore, atLeast(2)).put(captor.capture(), any());
+
+        JobStoreEntry last = captor.getValue();
+        assertEquals(jobId, last.jobId());
+        assertEquals(JobState.COMPLETE, last.state());
+        assertEquals("test-node", last.owningNodeId());
+    }
+
+    @Test
+    void testFindJobKeyByFileId_FallsBackToJobStore() {
+        // When the file id is not in the local map, TaskManager delegates to JobStore.
+        String fileId = "remote-file-id";
+        String expectedJobKey = "remote-job-key";
+        when(jobStore.findJobIdByFileId(fileId)).thenReturn(Optional.of(expectedJobKey));
+
+        String actual = taskManager.findJobKeyByFileId(fileId);
+
+        assertEquals(expectedJobKey, actual);
+        verify(jobStore).findJobIdByFileId(fileId);
+    }
 }

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

@@ -2,6 +2,8 @@ package stirling.software.common.util;
 
 import static org.junit.jupiter.api.Assertions.*;
 
+import java.nio.file.Files;
+import java.nio.file.Path;
 import java.util.ArrayList;
 import java.util.Collections;
 import java.util.List;
@@ -12,9 +14,47 @@ import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicInteger;
 
 import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.io.TempDir;
+import org.mockito.MockedStatic;
+import org.mockito.Mockito;
+
+import stirling.software.common.configuration.InstallationPathConfig;
 
 public class GeneralUtilsTest {
 
+    // Regression guard for the SSO auto-login persistence bug: the admin UI writes camelCase
+    // proFeatures keys, so saveKeyToSettings must match (and persist) them against the camelCase
+    // settings.yml.template. A case mismatch makes YamlHelper.updateValue silently no-op.
+    @Test
+    void saveKeyToSettings_persistsCamelCaseProFeatureKeys(@TempDir Path tempDir) throws Exception {
+        Path settings = tempDir.resolve("settings.yml");
+        Files.writeString(
+                settings,
+                """
+                premium:
+                  proFeatures:
+                    ssoAutoLogin: false
+                    customMetadata:
+                      author: username
+                """);
+
+        try (MockedStatic<InstallationPathConfig> mocked =
+                Mockito.mockStatic(InstallationPathConfig.class)) {
+            mocked.when(InstallationPathConfig::getSettingsPath).thenReturn(settings.toString());
+
+            GeneralUtils.saveKeyToSettings("premium.proFeatures.ssoAutoLogin", true);
+            GeneralUtils.saveKeyToSettings("premium.proFeatures.customMetadata.author", "alice");
+        }
+
+        YamlHelper reloaded = new YamlHelper(settings);
+        assertEquals(
+                "true", reloaded.getValueByExactKeyPath("premium", "proFeatures", "ssoAutoLogin"));
+        assertEquals(
+                "alice",
+                reloaded.getValueByExactKeyPath(
+                        "premium", "proFeatures", "customMetadata", "author"));
+    }
+
     @Test
     void testParsePageListWithAll() {
         List<Integer> result = GeneralUtils.parsePageList(new String[] {"all"}, 5, false);

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

@@ -98,6 +98,17 @@ class RequestUriUtilsTest {
         assertTrue(RequestUriUtils.isFrontendRoute("", "/split-pdf"));
     }
 
+    @Test
+    void testIsFrontendRoute_filesRouteOwnedByFrontend() {
+        // /files and /files/<folder-uuid> are FileManagerView routes - they
+        // must fall through to the SPA index.html, not get blocked by the
+        // backend auth filter. Regression test for direct-nav/refresh on
+        // the file manager returning a 401 JSON.
+        assertTrue(RequestUriUtils.isFrontendRoute("", "/files"));
+        assertTrue(
+                RequestUriUtils.isFrontendRoute("", "/files/3331910a-4155-4f71-8111-e38c896bc458"));
+    }
+
     @Test
     void testIsFrontendRoute_pathWithExtension() {
         assertFalse(RequestUriUtils.isFrontendRoute("", "/some/file.pdf"));
@@ -183,7 +194,7 @@ class RequestUriUtilsTest {
 
     @Test
     void testIsPublicAuthEndpoint_shareRootNotPublic() {
-        // Avoid matching bare "/share" or "/share/" — must have a token segment
+        // Avoid matching bare "/share" or "/share/" - must have a token segment
         assertFalse(RequestUriUtils.isPublicAuthEndpoint("/share", ""));
         assertFalse(RequestUriUtils.isPublicAuthEndpoint("/share/", ""));
     }
@@ -197,7 +208,7 @@ class RequestUriUtilsTest {
 
     @Test
     void testIsPublicAuthEndpoint_shareApiStillProtected() {
-        // Share-link data APIs must NOT be public — they enforce auth + access checks
+        // Share-link data APIs must NOT be public - they enforce auth + access checks
         assertFalse(RequestUriUtils.isPublicAuthEndpoint("/api/v1/storage/share-links/abc123", ""));
         assertFalse(
                 RequestUriUtils.isPublicAuthEndpoint(

app/common/src/test/java/stirling/software/common/util/propertyeditor/StringToArrayListPropertyEditorTest.java

@@ -1,163 +0,0 @@
-package stirling.software.common.util.propertyeditor;
-
-import static org.junit.jupiter.api.Assertions.*;
-
-import java.util.List;
-
-import org.junit.jupiter.api.BeforeEach;
-import org.junit.jupiter.api.Test;
-
-import stirling.software.common.model.api.security.RedactionArea;
-
-class StringToArrayListPropertyEditorTest {
-
-    private StringToArrayListPropertyEditor<RedactionArea> editor;
-
-    @BeforeEach
-    void setUp() {
-        editor = new StringToArrayListPropertyEditor<>(RedactionArea.class);
-    }
-
-    @Test
-    void testSetAsText_ValidJson() {
-        // Arrange
-        String json =
-                "[{\"x\":10.5,\"y\":20.5,\"width\":100.0,\"height\":50.0,\"page\":1,\"color\":\"#FF0000\"}]";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(1, list.size(), "List should have 1 entry");
-
-        RedactionArea area = list.get(0);
-        assertEquals(10.5, area.getX(), "X should be 10.5");
-        assertEquals(20.5, area.getY(), "Y should be 20.5");
-        assertEquals(100.0, area.getWidth(), "Width should be 100.0");
-        assertEquals(50.0, area.getHeight(), "Height should be 50.0");
-        assertEquals(1, area.getPage(), "Page should be 1");
-        assertEquals("#FF0000", area.getColor(), "Color should be #FF0000");
-    }
-
-    @Test
-    void testSetAsText_MultipleItems() {
-        // Arrange
-        String json =
-                "["
-                        + "{\"x\":10.0,\"y\":20.0,\"width\":100.0,\"height\":50.0,\"page\":1,\"color\":\"#FF0000\"},"
-                        + "{\"x\":30.0,\"y\":40.0,\"width\":200.0,\"height\":150.0,\"page\":2,\"color\":\"#00FF00\"}"
-                        + "]";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(2, list.size(), "List should have 2 entries");
-
-        RedactionArea area1 = list.get(0);
-        assertEquals(10.0, area1.getX(), "X should be 10.0");
-        assertEquals(20.0, area1.getY(), "Y should be 20.0");
-        assertEquals(1, area1.getPage(), "Page should be 1");
-
-        RedactionArea area2 = list.get(1);
-        assertEquals(30.0, area2.getX(), "X should be 30.0");
-        assertEquals(40.0, area2.getY(), "Y should be 40.0");
-        assertEquals(2, area2.getPage(), "Page should be 2");
-    }
-
-    @Test
-    void testSetAsText_EmptyString() {
-        // Arrange
-        String json = "";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertTrue(list.isEmpty(), "List should be empty");
-    }
-
-    @Test
-    void testSetAsText_NullString() {
-        // Act
-        editor.setAsText(null);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertTrue(list.isEmpty(), "List should be empty");
-    }
-
-    @Test
-    void testSetAsText_SingleItemAsArray() {
-        // Arrange - note this is a single object, not an array
-        String json =
-                "{\"x\":10.0,\"y\":20.0,\"width\":100.0,\"height\":50.0,\"page\":1,\"color\":\"#FF0000\"}";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(1, list.size(), "List should have 1 entry");
-
-        RedactionArea area = list.get(0);
-        assertEquals(10.0, area.getX(), "X should be 10.0");
-        assertEquals(20.0, area.getY(), "Y should be 20.0");
-    }
-
-    @Test
-    void testSetAsText_InvalidJson() {
-        // Arrange
-        String json = "invalid json";
-
-        // Act & Assert
-        assertThrows(IllegalArgumentException.class, () -> editor.setAsText(json));
-    }
-
-    @Test
-    void testSetAsText_UnknownProperties() {
-        // Arrange - this JSON contains properties not in RedactionArea
-        // With FAIL_ON_UNKNOWN_PROPERTIES disabled, this should ignore the unknown properties
-        String json = "[{\"invalid\":\"structure\"}]";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(1, list.size(), "List should have 1 entry (empty object)");
-    }
-}

app/core/build.gradle

@@ -207,13 +207,29 @@ def resourcesStaticDir = file('src/main/resources/static')
 def generatedFrontendPaths = [
     'assets',
     'index.html',
+    'index.html.gz',
+    'index.html.br',
+    'sw.js',
+    'sw.js.gz',
+    'sw.js.br',
+    'manifest.json.gz',
+    'manifest.json.br',
+    'site.webmanifest.gz',
+    'site.webmanifest.br',
+    'browserconfig.xml.gz',
+    'browserconfig.xml.br',
+    'manifest-classic.json',
+    'manifest-classic.json.gz',
+    'manifest-classic.json.br',
     'locales',
     'Login',
     'classic-logo',
     'modern-logo',
     'og_images',
     'samples',
-    'manifest-classic.json'
+    'pdfium',
+    'vendor',
+    'pdfjs'
 ]
 
 tasks.register('npmInstall', Exec) {

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

@@ -21,6 +21,13 @@ public class EndpointInterceptor implements HandlerInterceptor {
             HttpServletRequest request, HttpServletResponse response, Object handler)
             throws Exception {
         String requestURI = request.getRequestURI();
+
+        // Prevent API responses from being stored by browsers or intermediary caches by default
+        String servletPath = request.getServletPath();
+        if (servletPath != null && servletPath.startsWith("/api/")) {
+            response.setHeader("Cache-Control", "private, no-store");
+        }
+
         boolean isEnabled = endpointConfiguration.isEndpointEnabledForUri(requestURI);
         if (!isEnabled) {
             response.sendError(HttpServletResponse.SC_FORBIDDEN, "This endpoint is disabled");

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

@@ -1,5 +1,8 @@
 package stirling.software.SPDF.config;
 
+import java.time.Duration;
+import java.util.ArrayList;
+import java.util.List;
 import java.util.concurrent.TimeUnit;
 
 import org.slf4j.Logger;
@@ -10,6 +13,7 @@ import org.springframework.web.servlet.config.annotation.CorsRegistry;
 import org.springframework.web.servlet.config.annotation.InterceptorRegistry;
 import org.springframework.web.servlet.config.annotation.ResourceHandlerRegistry;
 import org.springframework.web.servlet.config.annotation.WebMvcConfigurer;
+import org.springframework.web.servlet.resource.EncodedResourceResolver;
 
 import lombok.RequiredArgsConstructor;
 
@@ -24,6 +28,10 @@ public class WebMvcConfig implements WebMvcConfigurer {
 
     private static final Logger logger = LoggerFactory.getLogger(WebMvcConfig.class);
 
+    private static final CacheControl NO_CACHE = CacheControl.noCache();
+    private static final CacheControl IMMUTABLE_ONE_YEAR =
+            CacheControl.maxAge(365, TimeUnit.DAYS).cachePublic().immutable();
+
     @Override
     public void addInterceptors(InterceptorRegistry registry) {
         registry.addInterceptor(endpointInterceptor);
@@ -31,37 +39,106 @@ public class WebMvcConfig implements WebMvcConfigurer {
 
     @Override
     public void addResourceHandlers(ResourceHandlerRegistry registry) {
-        // Cache hashed assets (JS/CSS with content hashes) for 1 year
-        // These files have names like index-ChAS4tCC.js that change when content changes
-        // Check customFiles/static first, then fall back to classpath
+        String staticPath =
+                "file:"
+                        + stirling.software.common.configuration.InstallationPathConfig
+                                .getStaticPath();
+
+        // 1. Service worker and PWA metadata (never store)
+        // Browsers revalidate SW bytes anyway; no-store is the safest for atomic updates.
+        registry.addResourceHandler(
+                        "/sw.js", "/manifest.json", "/site.webmanifest", "/browserconfig.xml")
+                .addResourceLocations(staticPath, "classpath:/static/")
+                .setCacheControl(CacheControl.noStore())
+                .resourceChain(true)
+                .addResolver(new EncodedResourceResolver());
+
+        // 2. Vite fingerprinted assets (immutable)
+        // These already have content hashes in filenames (e.g. index-ChAS4tCC.js)
         registry.addResourceHandler("/assets/**")
-                .addResourceLocations(
-                        "file:"
-                                + stirling.software.common.configuration.InstallationPathConfig
-                                        .getStaticPath()
-                                + "assets/",
-                        "classpath:/static/assets/")
-                .setCacheControl(CacheControl.maxAge(365, TimeUnit.DAYS).cachePublic());
+                .addResourceLocations(staticPath + "assets/", "classpath:/static/assets/")
+                .setCacheControl(IMMUTABLE_ONE_YEAR)
+                .resourceChain(true)
+                .addResolver(new EncodedResourceResolver());
 
-        // Don't cache index.html - it needs to be fresh to reference latest hashed assets
-        // Note: index.html is handled by ReactRoutingController for dynamic processing
-        registry.addResourceHandler("/index.html")
+        // 3. Media and fonts (immutable)
+        registry.addResourceHandler("/images/**", "/fonts/**")
                 .addResourceLocations(
-                        "file:"
-                                + stirling.software.common.configuration.InstallationPathConfig
-                                        .getStaticPath(),
-                        "classpath:/static/")
-                .setCacheControl(CacheControl.noCache().mustRevalidate());
+                        staticPath + "images/",
+                        "classpath:/static/images/",
+                        staticPath + "fonts/",
+                        "classpath:/static/fonts/")
+                .setCacheControl(IMMUTABLE_ONE_YEAR)
+                .resourceChain(true)
+                .addResolver(new EncodedResourceResolver());
 
-        // Handle all other static resources (js, css, images, fonts, etc.)
-        // Check customFiles/static first for user overrides
+        // 4. Branding and stable non-fingerprinted assets (1 day + SWR)
+        // Use stale-while-revalidate to improve perceived performance.
+        registry.addResourceHandler(
+                        "/favicon.*",
+                        "/apple-touch-icon.png",
+                        "/android-chrome-*.png",
+                        "/mstile-*.png",
+                        "/safari-pinned-tab.svg",
+                        "/icons/**",
+                        "/modern-logo/**",
+                        "/classic-logo/**",
+                        "/robots.txt",
+                        "/3rdPartyLicenses.json",
+                        "/pdfjs/**",
+                        "/pdfjs-legacy/**",
+                        "/pdfium/**",
+                        "/locales/**",
+                        "/css/**",
+                        "/js/**",
+                        "/vendor/**",
+                        "/samples/**",
+                        "/og_images/**",
+                        "/Login/**",
+                        "/manifest-classic.json")
+                .addResourceLocations(
+                        staticPath,
+                        "classpath:/static/",
+                        staticPath + "pdfjs/",
+                        "classpath:/static/pdfjs/",
+                        staticPath + "pdfjs-legacy/",
+                        "classpath:/static/pdfjs-legacy/",
+                        staticPath + "pdfium/",
+                        "classpath:/static/pdfium/",
+                        staticPath + "locales/",
+                        "classpath:/static/locales/",
+                        staticPath + "css/",
+                        "classpath:/static/css/",
+                        staticPath + "js/",
+                        "classpath:/static/js/",
+                        staticPath + "vendor/",
+                        "classpath:/static/vendor/",
+                        staticPath + "samples/",
+                        "classpath:/static/samples/",
+                        staticPath + "og_images/",
+                        "classpath:/static/og_images/",
+                        staticPath + "Login/",
+                        "classpath:/static/Login/",
+                        staticPath + "icons/",
+                        "classpath:/static/icons/",
+                        staticPath + "modern-logo/",
+                        "classpath:/static/modern-logo/",
+                        staticPath + "classic-logo/",
+                        "classpath:/static/classic-logo/")
+                .setCacheControl(
+                        CacheControl.maxAge(Duration.ofDays(1))
+                                .cachePublic()
+                                .staleWhileRevalidate(Duration.ofDays(7)))
+                .resourceChain(true)
+                .addResolver(new EncodedResourceResolver());
+
+        // 5. Catch-all (SPA fallback)
+        // Must check with server to ensure index.html is always fresh.
         registry.addResourceHandler("/**")
-                .addResourceLocations(
-                        "file:"
-                                + stirling.software.common.configuration.InstallationPathConfig
-                                        .getStaticPath(),
-                        "classpath:/static/")
-                .setCacheControl(CacheControl.maxAge(1, TimeUnit.HOURS));
+                .addResourceLocations(staticPath, "classpath:/static/")
+                .setCacheControl(NO_CACHE)
+                .resourceChain(true)
+                .addResolver(new EncodedResourceResolver());
     }
 
     @Override
@@ -115,9 +192,8 @@ public class WebMvcConfig implements WebMvcConfigurer {
                     applicationProperties.getSystem().getCorsAllowedOrigins());
 
             // Combine user-configured origins with Tauri origins
-            java.util.List<String> allOrigins =
-                    new java.util.ArrayList<>(
-                            applicationProperties.getSystem().getCorsAllowedOrigins());
+            List<String> allOrigins =
+                    new ArrayList<>(applicationProperties.getSystem().getCorsAllowedOrigins());
 
             // Always include Tauri origins for desktop app compatibility
             // Tauri v1 uses tauri://localhost, v2 uses http(s)://tauri.localhost
@@ -158,7 +234,8 @@ public class WebMvcConfig implements WebMvcConfigurer {
         } else {
             // Default to allowing all origins when nothing is configured
             logger.debug(
-                    "No CORS allowed origins configured in settings.yml (system.corsAllowedOrigins); WebMvcConfig allowing all origins.");
+                    "No CORS allowed origins configured in settings.yml"
+                            + " (system.corsAllowedOrigins); WebMvcConfig allowing all origins.");
             registry.addMapping("/**")
                     .allowedOriginPatterns("*")
                     .allowedMethods("GET", "POST", "PUT", "PATCH", "DELETE", "OPTIONS")

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

@@ -32,13 +32,16 @@ import stirling.software.SPDF.model.json.PdfJsonTextElement;
 import stirling.software.SPDF.service.PdfJsonConversionService;
 import stirling.software.common.annotations.AutoJobPostMapping;
 import stirling.software.common.annotations.api.GeneralApi;
+import stirling.software.common.enumeration.ResourceWeight;
 import stirling.software.common.model.api.general.EditTextOperation;
 import stirling.software.common.util.ExceptionUtils;
 import stirling.software.common.util.GeneralUtils;
 import stirling.software.common.util.TempFile;
 import stirling.software.common.util.TempFileManager;
 import stirling.software.common.util.WebResponseUtils;
-import stirling.software.common.util.propertyeditor.StringToArrayListPropertyEditor;
+import stirling.software.common.util.propertyeditor.JsonListPropertyEditor;
+
+import tools.jackson.core.type.TypeReference;
 
 /**
  * Find/replace text editing for PDFs. Round-trips through {@link PdfJsonConversionService}: the
@@ -72,10 +75,13 @@ public class EditTextController {
         binder.registerCustomEditor(
                 List.class,
                 "edits",
-                new StringToArrayListPropertyEditor<>(EditTextOperation.class));
+                new JsonListPropertyEditor<>(new TypeReference<List<EditTextOperation>>() {}));
     }
 
-    @AutoJobPostMapping(consumes = "multipart/form-data", value = "/edit-text")
+    @AutoJobPostMapping(
+            consumes = "multipart/form-data",
+            value = "/edit-text",
+            resourceWeight = ResourceWeight.LARGE_WEIGHT)
     @StandardPdfResponse
     @Operation(
             summary = "Edit text in a PDF via find and replace",

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

@@ -40,7 +40,8 @@ public class ScalePagesController {
     private final CustomPDFDocumentFactory pdfDocumentFactory;
     private final TempFileManager tempFileManager;
 
-    private static PDRectangle getTargetSize(String targetPDRectangle, PDDocument sourceDocument) {
+    private static PDRectangle getTargetSize(
+            String targetPDRectangle, String orientation, PDDocument sourceDocument) {
         if ("KEEP".equals(targetPDRectangle)) {
             if (sourceDocument.getNumberOfPages() == 0) {
                 throw ExceptionUtils.createInvalidPageSizeException("KEEP");
@@ -57,18 +58,19 @@ public class ScalePagesController {
         }
 
         Map<String, PDRectangle> sizeMap = getSizeMap();
-
-        if (sizeMap.containsKey(targetPDRectangle)) {
-            return sizeMap.get(targetPDRectangle);
+        PDRectangle base = sizeMap.get(targetPDRectangle);
+        if (base == null) {
+            throw ExceptionUtils.createInvalidPageSizeException(targetPDRectangle);
         }
 
-        throw ExceptionUtils.createInvalidPageSizeException(targetPDRectangle);
+        if ("LANDSCAPE".equalsIgnoreCase(orientation)) {
+            return new PDRectangle(base.getHeight(), base.getWidth());
+        }
+        return base;
     }
 
     private static Map<String, PDRectangle> getSizeMap() {
         Map<String, PDRectangle> sizeMap = new HashMap<>();
-
-        // Portrait sizes (A0-A6)
         sizeMap.put("A0", PDRectangle.A0);
         sizeMap.put("A1", PDRectangle.A1);
         sizeMap.put("A2", PDRectangle.A2);
@@ -76,42 +78,8 @@ public class ScalePagesController {
         sizeMap.put("A4", PDRectangle.A4);
         sizeMap.put("A5", PDRectangle.A5);
         sizeMap.put("A6", PDRectangle.A6);
-
-        // Landscape sizes (A0-A6)
-        sizeMap.put(
-                "A0_LANDSCAPE",
-                new PDRectangle(PDRectangle.A0.getHeight(), PDRectangle.A0.getWidth()));
-        sizeMap.put(
-                "A1_LANDSCAPE",
-                new PDRectangle(PDRectangle.A1.getHeight(), PDRectangle.A1.getWidth()));
-        sizeMap.put(
-                "A2_LANDSCAPE",
-                new PDRectangle(PDRectangle.A2.getHeight(), PDRectangle.A2.getWidth()));
-        sizeMap.put(
-                "A3_LANDSCAPE",
-                new PDRectangle(PDRectangle.A3.getHeight(), PDRectangle.A3.getWidth()));
-        sizeMap.put(
-                "A4_LANDSCAPE",
-                new PDRectangle(PDRectangle.A4.getHeight(), PDRectangle.A4.getWidth()));
-        sizeMap.put(
-                "A5_LANDSCAPE",
-                new PDRectangle(PDRectangle.A5.getHeight(), PDRectangle.A5.getWidth()));
-        sizeMap.put(
-                "A6_LANDSCAPE",
-                new PDRectangle(PDRectangle.A6.getHeight(), PDRectangle.A6.getWidth()));
-
-        // Portrait US sizes
         sizeMap.put("LETTER", PDRectangle.LETTER);
         sizeMap.put("LEGAL", PDRectangle.LEGAL);
-
-        // Landscape US sizes
-        sizeMap.put(
-                "LETTER_LANDSCAPE",
-                new PDRectangle(PDRectangle.LETTER.getHeight(), PDRectangle.LETTER.getWidth()));
-        sizeMap.put(
-                "LEGAL_LANDSCAPE",
-                new PDRectangle(PDRectangle.LEGAL.getHeight(), PDRectangle.LEGAL.getWidth()));
-
         return sizeMap;
     }
 
@@ -128,13 +96,14 @@ public class ScalePagesController {
             throws IOException {
         MultipartFile file = request.getFileInput();
         String targetPDRectangle = request.getPageSize();
+        String orientation = request.getOrientation();
         float scaleFactor = request.getScaleFactor();
 
         try (PDDocument sourceDocument = pdfDocumentFactory.load(file);
                 PDDocument outputDocument =
                         pdfDocumentFactory.createNewDocumentBasedOnOldDocument(sourceDocument)) {
 
-            PDRectangle targetSize = getTargetSize(targetPDRectangle, sourceDocument);
+            PDRectangle targetSize = getTargetSize(targetPDRectangle, orientation, sourceDocument);
 
             // Create LayerUtility once outside the loop for better performance
             LayerUtility layerUtility = new LayerUtility(outputDocument);

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

@@ -275,7 +275,10 @@ public class ConvertImgPDFController {
                 GeneralUtils.generateFilename(file[0].getOriginalFilename(), "_converted.pdf"));
     }
 
-    @AutoJobPostMapping(consumes = MediaType.MULTIPART_FORM_DATA_VALUE, value = "/cbz/pdf")
+    @AutoJobPostMapping(
+            consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
+            value = "/cbz/pdf",
+            resourceWeight = ResourceWeight.MEDIUM_WEIGHT)
     @Operation(
             summary = "Convert CBZ comic book archive to PDF",
             description =
@@ -301,7 +304,10 @@ public class ConvertImgPDFController {
         return WebResponseUtils.pdfFileToWebResponse(pdfFile, filename);
     }
 
-    @AutoJobPostMapping(consumes = MediaType.MULTIPART_FORM_DATA_VALUE, value = "/pdf/cbz")
+    @AutoJobPostMapping(
+            consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
+            value = "/pdf/cbz",
+            resourceWeight = ResourceWeight.LARGE_WEIGHT)
     @Operation(
             summary = "Convert PDF to CBZ comic book archive",
             description =
@@ -324,7 +330,10 @@ public class ConvertImgPDFController {
         return WebResponseUtils.zipFileToWebResponse(cbzFile, filename);
     }
 
-    @AutoJobPostMapping(consumes = MediaType.MULTIPART_FORM_DATA_VALUE, value = "/cbr/pdf")
+    @AutoJobPostMapping(
+            consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
+            value = "/cbr/pdf",
+            resourceWeight = ResourceWeight.MEDIUM_WEIGHT)
     @Operation(
             summary = "Convert CBR comic book archive to PDF",
             description =
@@ -350,7 +359,10 @@ public class ConvertImgPDFController {
         return WebResponseUtils.bytesToWebResponse(pdfBytes, filename);
     }
 
-    @AutoJobPostMapping(consumes = MediaType.MULTIPART_FORM_DATA_VALUE, value = "/pdf/cbr")
+    @AutoJobPostMapping(
+            consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
+            value = "/pdf/cbr",
+            resourceWeight = ResourceWeight.LARGE_WEIGHT)
     @Operation(
             summary = "Convert PDF to CBR comic book archive",
             description =

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

@@ -141,7 +141,8 @@ public class AttachmentController {
 
     @AutoJobPostMapping(
             consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
-            value = "/extract-attachments")
+            value = "/extract-attachments",
+            resourceWeight = ResourceWeight.SMALL_WEIGHT)
     @Operation(
             summary = "Extract attachments from PDF",
             description =
@@ -176,7 +177,10 @@ public class AttachmentController {
         }
     }
 
-    @AutoJobPostMapping(consumes = MediaType.MULTIPART_FORM_DATA_VALUE, value = "/list-attachments")
+    @AutoJobPostMapping(
+            consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
+            value = "/list-attachments",
+            resourceWeight = ResourceWeight.SMALL_WEIGHT)
     @Operation(
             summary = "List attachments in PDF",
             description =
@@ -193,7 +197,8 @@ public class AttachmentController {
 
     @AutoJobPostMapping(
             consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
-            value = "/rename-attachment")
+            value = "/rename-attachment",
+            resourceWeight = ResourceWeight.SMALL_WEIGHT)
     @StandardPdfResponse
     @Operation(
             summary = "Rename attachment in PDF",
@@ -228,7 +233,8 @@ public class AttachmentController {
 
     @AutoJobPostMapping(
             consumes = MediaType.MULTIPART_FORM_DATA_VALUE,
-            value = "/delete-attachment")
+            value = "/delete-attachment",
+            resourceWeight = ResourceWeight.SMALL_WEIGHT)
     @StandardPdfResponse
     @Operation(
             summary = "Delete attachment from PDF",

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

@@ -12,6 +12,8 @@ import org.springframework.web.bind.annotation.RequestParam;
 
 import io.swagger.v3.oas.annotations.Hidden;
 
+import jakarta.servlet.http.HttpServletRequest;
+
 import lombok.extern.slf4j.Slf4j;
 
 import stirling.software.SPDF.config.EndpointConfiguration;
@@ -91,6 +93,63 @@ public class ConfigController {
         return null;
     }
 
+    /**
+     * Resolve the frontend URL the client should advertise to phones / share-link recipients.
+     * Priority: explicit system.frontendUrl, then the Host the user is already using to reach this
+     * server (works for Docker, reverse proxies, and bare-metal LANs), then a detected site-local
+     * IPv4, then empty.
+     */
+    // visible for testing
+    String resolveFrontendUrl(HttpServletRequest request, AppConfig appConfig) {
+        String configured = applicationProperties.getSystem().getFrontendUrl();
+        if (configured != null && !configured.isBlank()) {
+            return configured;
+        }
+        if (request != null) {
+            String host = request.getServerName();
+            if (host != null && !host.isBlank() && !isLoopbackHost(host)) {
+                String scheme = request.getScheme();
+                int port = request.getServerPort();
+                boolean defaultPort =
+                        ("http".equals(scheme) && port == 80)
+                                || ("https".equals(scheme) && port == 443);
+                return defaultPort ? scheme + "://" + host : scheme + "://" + host + ":" + port;
+            }
+        }
+        String localIp = GeneralUtils.getLocalNetworkIp();
+        if (localIp != null) {
+            String scheme = appConfig.getBackendUrl().startsWith("https") ? "https" : "http";
+            return scheme + "://" + localIp + ":" + resolveEffectiveServerPort(appConfig);
+        }
+        return "";
+    }
+
+    /**
+     * The port the embedded server is actually listening on. With {@code server.port=0} (an
+     * ephemeral port, which the desktop bundle uses to dodge port clashes) the configured value
+     * stays {@code "0"} while Spring publishes the real bound port as {@code local.server.port}
+     * once the server is up. Advertised URLs (the mobile-scanner QR, share links) must carry the
+     * real port - a literal {@code :0} is unreachable and browsers reject it as ERR_UNSAFE_PORT.
+     */
+    // visible for testing
+    String resolveEffectiveServerPort(AppConfig appConfig) {
+        String configured = appConfig.getServerPort();
+        if (configured == null || "0".equals(configured.trim())) {
+            String actual = applicationContext.getEnvironment().getProperty("local.server.port");
+            if (actual != null && !actual.isBlank()) {
+                return actual;
+            }
+        }
+        return configured;
+    }
+
+    private static boolean isLoopbackHost(String host) {
+        return "localhost".equalsIgnoreCase(host)
+                || "127.0.0.1".equals(host)
+                || "::1".equals(host)
+                || "0:0:0:0:0:0:0:1".equals(host);
+    }
+
     /** Check if running Enterprise edition dynamically. */
     private Boolean isRunningEE() {
         // Use LicenseService for fresh license status if available
@@ -107,7 +166,7 @@ public class ConfigController {
     }
 
     @GetMapping("/app-config")
-    public ResponseEntity<Map<String, Object>> getAppConfig() {
+    public ResponseEntity<Map<String, Object>> getAppConfig(HttpServletRequest request) {
         Map<String, Object> configData = new HashMap<>();
 
         try {
@@ -121,20 +180,10 @@ public class ConfigController {
             // Note: Frontend expects "baseUrl" field name for compatibility
             configData.put("baseUrl", appConfig.getBackendUrl());
             configData.put("contextPath", appConfig.getContextPath());
-            configData.put("serverPort", appConfig.getServerPort());
+            configData.put("serverPort", resolveEffectiveServerPort(appConfig));
 
             String frontendUrl = applicationProperties.getSystem().getFrontendUrl();
-            if ((frontendUrl == null || frontendUrl.isBlank())
-                    && Boolean.parseBoolean(
-                            System.getProperty("STIRLING_PDF_TAURI_MODE", "false"))) {
-                String localIp = GeneralUtils.getLocalNetworkIp();
-                if (localIp != null) {
-                    String scheme =
-                            appConfig.getBackendUrl().startsWith("https") ? "https" : "http";
-                    frontendUrl = scheme + "://" + localIp + ":" + appConfig.getServerPort();
-                }
-            }
-            configData.put("frontendUrl", frontendUrl != null ? frontendUrl : "");
+            configData.put("frontendUrl", resolveFrontendUrl(request, appConfig));
 
             // Add mobile scanner settings
             configData.put(
@@ -277,6 +326,9 @@ public class ConfigController {
             // Premium/Enterprise settings
             configData.put("premiumEnabled", applicationProperties.getPremium().isEnabled());
 
+            // AI Engine settings
+            configData.put("aiEngineEnabled", applicationProperties.getAiEngine().isEnabled());
+
             // Timestamp TSA settings — single source of truth for presets + admin URLs
             ApplicationProperties.Security.Timestamp tsConfig =
                     applicationProperties.getSecurity().getTimestamp();

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

@@ -0,0 +1,127 @@
+package stirling.software.SPDF.controller.api.security;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.text.PDFTextStripper;
+import org.apache.pdfbox.text.TextPosition;
+
+/**
+ * PDFTextStripper subclass that collects all text positions and groups them into line-level
+ * bounding boxes.
+ *
+ * <p>Two outputs are maintained in parallel:
+ *
+ * <ul>
+ *   <li>{@link #getLineBoxes()} returns {@code [x1, pdfYbottom, x2, pdfYtop]} in PDF user-space
+ *       (origin bottom-left, Y up). This is what existing callers expect.
+ *   <li>{@link #getScreenLineBoxes()} returns {@code [x1, screenYtop, x2, screenYbottom]} computed
+ *       directly from glyph positions without a PDF↔screen round-trip — used by column-aware
+ *       redaction where ulp-level drift in the round-trip caused false rejects against anchors.
+ * </ul>
+ *
+ * <p>Lines are flushed not only on Y jumps but also on large X gaps within the same Y row. That way
+ * left-column glyphs and right-column glyphs that happen to share a baseline (common in IEEE
+ * conference templates) get emitted as two distinct line boxes instead of one wide merged box.
+ */
+final class AllTextLineExtractor extends PDFTextStripper {
+
+    /** Min vertical jump (screen Y) before the next glyph is treated as a new line. */
+    private static final float LINE_Y_TOLERANCE = 3.0f;
+
+    /**
+     * Min horizontal gap (screen X) between consecutive glyphs on the same Y that indicates a
+     * column boundary. Chosen large enough to not split normal inter-word spacing (~6–10pt for 11pt
+     * text) but small enough to catch standard column gutters (typically ≥15pt).
+     */
+    private static final float COLUMN_GAP_X = 14f;
+
+    private final float pageHeight;
+    private final List<float[]> lineBoxes = new ArrayList<>();
+    private final List<float[]> screenLineBoxes = new ArrayList<>();
+
+    private final List<TextPosition> currentLine = new ArrayList<>();
+    private float lastScreenY = Float.NaN;
+    private float lastGlyphRight = Float.NaN;
+
+    AllTextLineExtractor(int pageNumber, float pageHeight) throws IOException {
+        this.pageHeight = pageHeight;
+        setStartPage(pageNumber);
+        setEndPage(pageNumber);
+        setSortByPosition(true);
+    }
+
+    List<float[]> getLineBoxes() {
+        return lineBoxes;
+    }
+
+    /**
+     * Returns line boxes as {@code [x1, screenYtop, x2, screenYbottom]}. {@code screenYtop} is the
+     * minimum {@code TextPosition.getY() - getHeight()} on the line and {@code screenYbottom} is
+     * the maximum {@code getY()} (the line's baseline). These values come straight from PDFBox
+     * without going through {@code pageHeight - …}, so they're stable for ulp-sensitive comparisons
+     * against anchor screen Ys.
+     */
+    List<float[]> getScreenLineBoxes() {
+        return screenLineBoxes;
+    }
+
+    @Override
+    protected void writeString(String text, List<TextPosition> positions) {
+        for (TextPosition tp : positions) {
+            // Skip whitespace-only positions (spaces, newline markers, indent characters).
+            // These have a TextPosition but no visible glyph; including them causes
+            // space-only "lines" to produce degenerate segments that appear as thin
+            // black bars after redaction.
+            String unicode = tp.getUnicode();
+            if (unicode == null || unicode.isBlank()) {
+                continue;
+            }
+            float screenY = tp.getY();
+            float screenX = tp.getX();
+            boolean yJump =
+                    !Float.isNaN(lastScreenY) && Math.abs(screenY - lastScreenY) > LINE_Y_TOLERANCE;
+            boolean xJump =
+                    !Float.isNaN(lastGlyphRight) && (screenX - lastGlyphRight) > COLUMN_GAP_X;
+            if (yJump || xJump) {
+                flushLine();
+            }
+            lastScreenY = screenY;
+            lastGlyphRight = screenX + tp.getWidth();
+            currentLine.add(tp);
+        }
+    }
+
+    @Override
+    protected void endPage(PDPage page) throws IOException {
+        flushLine();
+        super.endPage(page);
+    }
+
+    private void flushLine() {
+        if (currentLine.isEmpty()) {
+            return;
+        }
+        float minX = Float.MAX_VALUE, maxX = -Float.MAX_VALUE;
+        float minScreenY = Float.MAX_VALUE, maxScreenY = -Float.MAX_VALUE;
+        for (TextPosition tp : currentLine) {
+            minX = Math.min(minX, tp.getX());
+            maxX = Math.max(maxX, tp.getX() + tp.getWidth());
+            minScreenY = Math.min(minScreenY, tp.getY() - tp.getHeight());
+            maxScreenY = Math.max(maxScreenY, tp.getY());
+        }
+        emitSegment(minX, maxX, minScreenY, maxScreenY);
+        currentLine.clear();
+        lastScreenY = Float.NaN;
+        lastGlyphRight = Float.NaN;
+    }
+
+    private void emitSegment(float minX, float maxX, float minScreenY, float maxScreenY) {
+        float pdfY1 = pageHeight - maxScreenY; // bottom in PDF coords
+        float pdfY2 = pageHeight - minScreenY; // top in PDF coords
+        lineBoxes.add(new float[] {minX, pdfY1, maxX, pdfY2});
+        screenLineBoxes.add(new float[] {minX, minScreenY, maxX, maxScreenY});
+    }
+}

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

@@ -0,0 +1,403 @@
+package stirling.software.SPDF.controller.api.security;
+
+import java.awt.Color;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.PDPageContentStream;
+import org.apache.pdfbox.pdmodel.PDPageTree;
+import org.apache.pdfbox.pdmodel.common.PDRectangle;
+import org.apache.pdfbox.pdmodel.interactive.annotation.PDAnnotation;
+import org.springframework.stereotype.Service;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.SPDF.model.PDFText;
+import stirling.software.SPDF.model.api.security.ManualRedactPdfRequest;
+import stirling.software.SPDF.pdf.parser.PageImageLocator;
+import stirling.software.common.model.api.security.RedactionArea;
+import stirling.software.common.util.GeneralUtils;
+import stirling.software.common.util.PdfUtils;
+import stirling.software.common.util.TempFile;
+import stirling.software.common.util.TempFileManager;
+
+@Service
+@Slf4j
+@RequiredArgsConstructor
+class ManualRedactionService {
+
+    private static final float DEFAULT_TEXT_PADDING_MULTIPLIER = 0.6f;
+    private static final float REDACTION_WIDTH_REDUCTION_FACTOR = 0.9f;
+
+    private final TempFileManager tempFileManager;
+
+    // -----------------------------------------------------------------------
+    // Area and page redaction
+    // -----------------------------------------------------------------------
+
+    void redactAreas(List<RedactionArea> redactionAreas, PDDocument document, PDPageTree allPages)
+            throws IOException {
+
+        if (redactionAreas == null || redactionAreas.isEmpty()) {
+            return;
+        }
+
+        Map<Integer, List<RedactionArea>> redactionsByPage = new HashMap<>();
+
+        for (RedactionArea redactionArea : redactionAreas) {
+            if (redactionArea.getPage() == null
+                    || redactionArea.getPage() <= 0
+                    || redactionArea.getHeight() == null
+                    || redactionArea.getHeight() <= 0.0D
+                    || redactionArea.getWidth() == null
+                    || redactionArea.getWidth() <= 0.0D) {
+                continue;
+            }
+
+            redactionsByPage
+                    .computeIfAbsent(redactionArea.getPage(), k -> new ArrayList<>())
+                    .add(redactionArea);
+        }
+
+        for (Map.Entry<Integer, List<RedactionArea>> entry : redactionsByPage.entrySet()) {
+            Integer pageNumber = entry.getKey();
+            List<RedactionArea> areasForPage = entry.getValue();
+
+            if (pageNumber > allPages.getCount()) {
+                continue;
+            }
+
+            PDPage page = allPages.get(pageNumber - 1);
+
+            try (PDPageContentStream contentStream =
+                    new PDPageContentStream(
+                            document, page, PDPageContentStream.AppendMode.APPEND, true, true)) {
+
+                contentStream.saveGraphicsState();
+                for (RedactionArea redactionArea : areasForPage) {
+                    Color redactColor = decodeOrDefault(redactionArea.getColor());
+
+                    contentStream.setNonStrokingColor(redactColor);
+
+                    float x = redactionArea.getX().floatValue();
+                    float y = redactionArea.getY().floatValue();
+                    float width = redactionArea.getWidth().floatValue();
+                    float height = redactionArea.getHeight().floatValue();
+
+                    float pdfY = page.getBBox().getHeight() - y - height;
+
+                    contentStream.addRect(x, pdfY, width, height);
+                    contentStream.fill();
+                }
+                contentStream.restoreGraphicsState();
+            }
+        }
+    }
+
+    void redactPages(ManualRedactPdfRequest request, PDDocument document, PDPageTree allPages)
+            throws IOException {
+
+        Color redactColor = decodeOrDefault(request.getPageRedactionColor());
+        List<Integer> pageNumbers = getPageNumbers(request, allPages.getCount());
+
+        for (Integer pageNumber : pageNumbers) {
+            PDPage page = allPages.get(pageNumber);
+
+            try (PDPageContentStream contentStream =
+                    new PDPageContentStream(
+                            document, page, PDPageContentStream.AppendMode.APPEND, true, true)) {
+                contentStream.setNonStrokingColor(redactColor);
+
+                PDRectangle box = page.getBBox();
+                contentStream.addRect(0, 0, box.getWidth(), box.getHeight());
+                contentStream.fill();
+            }
+        }
+    }
+
+    // -----------------------------------------------------------------------
+    // Overlay drawing
+    // -----------------------------------------------------------------------
+
+    void redactFoundText(
+            PDDocument document,
+            List<PDFText> blocks,
+            float customPadding,
+            Color redactColor,
+            boolean isTextRemovalMode)
+            throws IOException {
+
+        var allPages = document.getDocumentCatalog().getPages();
+
+        Map<Integer, List<PDFText>> blocksByPage = new HashMap<>();
+        for (PDFText block : blocks) {
+            blocksByPage.computeIfAbsent(block.getPageIndex(), k -> new ArrayList<>()).add(block);
+        }
+
+        for (Map.Entry<Integer, List<PDFText>> entry : blocksByPage.entrySet()) {
+            Integer pageIndex = entry.getKey();
+            List<PDFText> pageBlocks = entry.getValue();
+
+            if (pageIndex >= allPages.getCount()) {
+                continue;
+            }
+
+            var page = allPages.get(pageIndex);
+            try (PDPageContentStream contentStream =
+                    new PDPageContentStream(
+                            document, page, PDPageContentStream.AppendMode.APPEND, true, true)) {
+
+                contentStream.saveGraphicsState();
+
+                try {
+                    contentStream.setNonStrokingColor(redactColor);
+                    PDRectangle pageBox = page.getBBox();
+
+                    for (PDFText block : pageBlocks) {
+                        float padding =
+                                (block.getY2() - block.getY1()) * DEFAULT_TEXT_PADDING_MULTIPLIER
+                                        + customPadding;
+
+                        float originalWidth = block.getX2() - block.getX1();
+                        float boxWidth;
+                        float boxX;
+
+                        if (isTextRemovalMode) {
+                            boxWidth = originalWidth * REDACTION_WIDTH_REDUCTION_FACTOR;
+                            float widthReduction = originalWidth - boxWidth;
+                            boxX = block.getX1() + (widthReduction / 2);
+                        } else {
+                            boxWidth = originalWidth;
+                            boxX = block.getX1();
+                        }
+
+                        contentStream.addRect(
+                                boxX,
+                                pageBox.getHeight() - block.getY2() - padding,
+                                boxWidth,
+                                block.getY2() - block.getY1() + 2 * padding);
+                    }
+
+                    contentStream.fill();
+
+                } finally {
+                    contentStream.restoreGraphicsState();
+                }
+            }
+
+            // Remove annotations whose bounding rect overlaps a redacted block, to prevent
+            // users from hovering over redacted URLs and seeing the underlying destination.
+            try {
+                float pageH = page.getBBox().getHeight();
+                List<PDAnnotation> kept = new ArrayList<>();
+                for (PDAnnotation ann : page.getAnnotations()) {
+                    PDRectangle ar = ann.getRectangle();
+                    boolean overlaps = false;
+                    if (ar != null) {
+                        for (PDFText block : pageBlocks) {
+                            float padding =
+                                    (block.getY2() - block.getY1())
+                                                    * DEFAULT_TEXT_PADDING_MULTIPLIER
+                                            + customPadding;
+                            float bx1 = block.getX1();
+                            float bx2 = block.getX2();
+                            float by1 = pageH - block.getY2() - padding;
+                            float by2 = pageH - block.getY1() + padding;
+                            if (ar.getLowerLeftX() < bx2
+                                    && ar.getUpperRightX() > bx1
+                                    && ar.getLowerLeftY() < by2
+                                    && ar.getUpperRightY() > by1) {
+                                overlaps = true;
+                                break;
+                            }
+                        }
+                    }
+                    if (!overlaps) {
+                        kept.add(ann);
+                    }
+                }
+                page.setAnnotations(kept);
+            } catch (Exception e) {
+                log.debug(
+                        "[redact] could not remove annotations on page {}: {}",
+                        pageIndex,
+                        e.getMessage());
+            }
+        }
+    }
+
+    void redactImageBoxes(PDDocument document, List<float[]> imageBoxes, Color color)
+            throws IOException {
+        Map<Integer, List<float[]>> byPage = new HashMap<>();
+        for (float[] box : imageBoxes) {
+            byPage.computeIfAbsent((int) box[0], k -> new ArrayList<>()).add(box);
+        }
+        PDPageTree pages = document.getDocumentCatalog().getPages();
+        for (Map.Entry<Integer, List<float[]>> entry : byPage.entrySet()) {
+            int pageIdx = entry.getKey();
+            if (pageIdx < 0 || pageIdx >= pages.getCount()) {
+                log.warn("[redact/execute] image box references out-of-range page {}", pageIdx);
+                continue;
+            }
+            PDPage page = pages.get(pageIdx);
+            try (PDPageContentStream cs =
+                    new PDPageContentStream(
+                            document, page, PDPageContentStream.AppendMode.APPEND, true, true)) {
+                cs.saveGraphicsState();
+                cs.setNonStrokingColor(color);
+                for (float[] box : entry.getValue()) {
+                    float x1 = box[1], y1 = box[2], x2 = box[3], y2 = box[4];
+                    cs.addRect(x1, y1, x2 - x1, y2 - y1);
+                }
+                cs.fill();
+                cs.restoreGraphicsState();
+            }
+        }
+    }
+
+    // -----------------------------------------------------------------------
+    // Page element extraction
+    // -----------------------------------------------------------------------
+
+    /**
+     * Returns bounding boxes for every text line and image on {@code page} in PDF user-space
+     * coordinates: {@code [x1, y1, x2, y2]} (origin bottom-left, Y increases upward).
+     */
+    List<float[]> extractPageElementBoxes(PDDocument document, PDPage page, int pageIndex)
+            throws IOException {
+        List<float[]> boxes = new ArrayList<>();
+
+        AllTextLineExtractor textExtractor =
+                new AllTextLineExtractor(pageIndex + 1, page.getBBox().getHeight());
+        textExtractor.getText(document);
+        boxes.addAll(textExtractor.getLineBoxes());
+
+        PageImageLocator imgLocator = new PageImageLocator(page, pageIndex);
+        imgLocator.processPage(page);
+        for (PageImageLocator.ImageBox imgBox : imgLocator.getImageBoxes()) {
+            boxes.add(new float[] {imgBox.x1(), imgBox.y1(), imgBox.x2(), imgBox.y2()});
+        }
+
+        return boxes;
+    }
+
+    // -----------------------------------------------------------------------
+    // Finalization
+    // -----------------------------------------------------------------------
+
+    TempFile finalizeRedaction(
+            PDDocument document,
+            Map<Integer, List<PDFText>> allFoundTextsByPage,
+            String colorString,
+            float customPadding,
+            Boolean convertToImage,
+            boolean isTextRemovalMode)
+            throws IOException {
+
+        List<PDFText> allFoundTexts = new ArrayList<>();
+        for (List<PDFText> pageTexts : allFoundTextsByPage.values()) {
+            allFoundTexts.addAll(pageTexts);
+        }
+
+        if (!allFoundTexts.isEmpty()) {
+            Color redactColor = decodeOrDefault(colorString);
+            redactFoundText(document, allFoundTexts, customPadding, redactColor, isTextRemovalMode);
+            cleanDocumentMetadata(document);
+        }
+
+        if (Boolean.TRUE.equals(convertToImage)) {
+            try (PDDocument convertedPdf = PdfUtils.convertPdfToPdfImage(document)) {
+                cleanDocumentMetadata(convertedPdf);
+
+                TempFile tempOut = tempFileManager.createManagedTempFile(".pdf");
+                try {
+                    convertedPdf.save(tempOut.getFile());
+                } catch (IOException e) {
+                    tempOut.close();
+                    throw e;
+                }
+
+                log.info(
+                        "Redaction finalized (image mode): {} pages ➜ {} KB",
+                        convertedPdf.getNumberOfPages(),
+                        tempOut.getFile().length() / 1024);
+
+                return tempOut;
+            }
+        }
+
+        TempFile tempOut = tempFileManager.createManagedTempFile(".pdf");
+        try {
+            document.save(tempOut.getFile());
+        } catch (IOException e) {
+            tempOut.close();
+            throw e;
+        }
+
+        log.info(
+                "Redaction finalized: {} pages ➜ {} KB",
+                document.getNumberOfPages(),
+                tempOut.getFile().length() / 1024);
+
+        return tempOut;
+    }
+
+    private void cleanDocumentMetadata(PDDocument document) {
+        try {
+            var documentInfo = document.getDocumentInformation();
+            if (documentInfo != null) {
+                documentInfo.setAuthor(null);
+                documentInfo.setSubject(null);
+                documentInfo.setKeywords(null);
+                documentInfo.setModificationDate(java.util.Calendar.getInstance());
+                log.debug("Cleaned document metadata for security");
+            }
+
+            if (document.getDocumentCatalog() != null) {
+                try {
+                    document.getDocumentCatalog().setMetadata(null);
+                } catch (Exception e) {
+                    log.debug("Could not clear XMP metadata: {}", e.getMessage());
+                }
+            }
+
+        } catch (Exception e) {
+            log.warn("Failed to clean document metadata: {}", e.getMessage());
+        }
+    }
+
+    // -----------------------------------------------------------------------
+    // Utilities
+    // -----------------------------------------------------------------------
+
+    static Color decodeOrDefault(String hex) {
+        if (hex == null) {
+            return Color.BLACK;
+        }
+
+        String colorString = hex.startsWith("#") ? hex : "#" + hex;
+
+        try {
+            return Color.decode(colorString);
+        } catch (NumberFormatException e) {
+            return Color.BLACK;
+        }
+    }
+
+    private List<Integer> getPageNumbers(ManualRedactPdfRequest request, int pagesCount) {
+        String pageNumbersInput = request.getPageNumbers();
+        String[] parsedPageNumbers =
+                pageNumbersInput != null ? pageNumbersInput.split(",") : new String[0];
+        List<Integer> pageNumbers =
+                GeneralUtils.parsePageList(parsedPageNumbers, pagesCount, false);
+        Collections.sort(pageNumbers);
+        return pageNumbers;
+    }
+}

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

@@ -0,0 +1,174 @@
+package stirling.software.SPDF.controller.api.security;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.ExecutorService;
+import java.util.concurrent.Executors;
+import java.util.concurrent.Future;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.TimeoutException;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.text.PDFTextStripper;
+import org.apache.pdfbox.text.TextPosition;
+
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.SPDF.model.PDFText;
+
+/**
+ * Scans a PDF document once and matches all provided patterns in a single pass, collecting
+ * bounding-box positions for every match. Use in place of creating one {@code TextFinder} per
+ * search term to avoid O(n) full-document scans.
+ */
+@Slf4j
+final class MultiPatternTextFinder extends PDFTextStripper {
+
+    private static final long REGEX_MATCH_TIMEOUT_SECONDS = 30;
+    private static final ExecutorService REGEX_EXECUTOR =
+            Executors.newVirtualThreadPerTaskExecutor();
+
+    private final List<Pattern> patterns;
+    private final Map<Integer, List<PDFText>> foundTextsByPage = new HashMap<>();
+
+    private final List<TextPosition> pageTextPositions = new ArrayList<>();
+    private final StringBuilder pageTextBuilder = new StringBuilder();
+
+    MultiPatternTextFinder(List<Pattern> patterns) throws IOException {
+        this.patterns = patterns;
+        this.setWordSeparator(" ");
+        this.setLineSeparator("\n");
+    }
+
+    Map<Integer, List<PDFText>> getFoundTextsByPage() {
+        return foundTextsByPage;
+    }
+
+    @Override
+    protected void startPage(PDPage page) throws IOException {
+        super.startPage(page);
+        pageTextPositions.clear();
+        pageTextBuilder.setLength(0);
+    }
+
+    @Override
+    protected void writeString(String text, List<TextPosition> textPositions) {
+        pageTextBuilder.append(text);
+        pageTextPositions.addAll(textPositions);
+    }
+
+    @Override
+    protected void writeWordSeparator() {
+        pageTextBuilder.append(getWordSeparator());
+        pageTextPositions.add(null);
+    }
+
+    @Override
+    protected void writeLineSeparator() {
+        pageTextBuilder.append(getLineSeparator());
+        pageTextPositions.add(null);
+    }
+
+    @Override
+    protected void endPage(PDPage page) throws IOException {
+        String text = pageTextBuilder.toString();
+        if (!text.isEmpty()) {
+            int pageIndex = getCurrentPageNo() - 1;
+            for (Pattern pattern : patterns) {
+                Matcher matcher = pattern.matcher(text);
+                while (safeFind(matcher)) {
+                    PDFText pdfText = resolveMatchPosition(matcher, pageIndex);
+                    if (pdfText != null) {
+                        foundTextsByPage
+                                .computeIfAbsent(pageIndex, k -> new ArrayList<>())
+                                .add(pdfText);
+                    }
+                }
+            }
+        }
+        super.endPage(page);
+    }
+
+    /**
+     * Wraps a single {@code matcher.find()} call with a {@value #REGEX_MATCH_TIMEOUT_SECONDS}
+     * second timeout. Prevents pathological regex backtracking from blocking the request
+     * indefinitely; per-match timeout so fast legitimate scans are unaffected.
+     */
+    private static boolean safeFind(Matcher matcher) throws IOException {
+        Future<Boolean> future =
+                REGEX_EXECUTOR.submit((java.util.concurrent.Callable<Boolean>) matcher::find);
+        try {
+            return future.get(REGEX_MATCH_TIMEOUT_SECONDS, TimeUnit.SECONDS);
+        } catch (TimeoutException e) {
+            future.cancel(true);
+            throw new IOException(
+                    "Regex match timed out after "
+                            + REGEX_MATCH_TIMEOUT_SECONDS
+                            + "s — pattern may cause catastrophic backtracking");
+        } catch (InterruptedException e) {
+            future.cancel(true);
+            Thread.currentThread().interrupt();
+            throw new IOException("Regex match interrupted", e);
+        } catch (ExecutionException e) {
+            Throwable cause = e.getCause();
+            if (cause instanceof IOException ioEx) throw ioEx;
+            throw new IOException("Regex match failed: " + cause.getMessage(), cause);
+        }
+    }
+
+    private PDFText resolveMatchPosition(Matcher matcher, int pageIndex) {
+        int matchStart = matcher.start();
+        int matchEnd = matcher.end();
+
+        float minX = Float.MAX_VALUE;
+        float minY = Float.MAX_VALUE;
+        float maxX = Float.MIN_VALUE;
+        float maxY = Float.MIN_VALUE;
+        boolean foundPosition = false;
+
+        for (int i = matchStart; i < matchEnd; i++) {
+            if (i >= pageTextPositions.size()) break;
+            TextPosition pos = pageTextPositions.get(i);
+            if (pos != null) {
+                foundPosition = true;
+                minX = Math.min(minX, pos.getX());
+                maxX = Math.max(maxX, pos.getX() + pos.getWidth());
+                minY = Math.min(minY, pos.getY() - pos.getHeight());
+                maxY = Math.max(maxY, pos.getY());
+            }
+        }
+
+        if (!foundPosition && matchStart < pageTextPositions.size()) {
+            for (int i = Math.max(0, matchStart - 5);
+                    i < Math.min(pageTextPositions.size(), matchEnd + 5);
+                    i++) {
+                TextPosition pos = pageTextPositions.get(i);
+                if (pos != null) {
+                    foundPosition = true;
+                    minX = Math.min(minX, pos.getX());
+                    maxX = Math.max(maxX, pos.getX() + pos.getWidth());
+                    minY = Math.min(minY, pos.getY() - pos.getHeight());
+                    maxY = Math.max(maxY, pos.getY());
+                    break;
+                }
+            }
+        }
+
+        if (!foundPosition) {
+            log.warn(
+                    "Found text match '{}' but no valid position data at {}-{}",
+                    matcher.group(),
+                    matchStart,
+                    matchEnd);
+            return null;
+        }
+
+        return new PDFText(pageIndex, minX, minY, maxX, maxY, matcher.group());
+    }
+}

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

@@ -0,0 +1,850 @@
+package stirling.software.SPDF.controller.api.security;
+
+import java.awt.Color;
+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.regex.Pattern;
+
+import org.apache.pdfbox.cos.COSName;
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.PDPageContentStream;
+import org.apache.pdfbox.pdmodel.PDPageTree;
+import org.apache.pdfbox.pdmodel.PDResources;
+import org.apache.pdfbox.pdmodel.common.PDRectangle;
+import org.springframework.stereotype.Service;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+import stirling.software.SPDF.model.PDFText;
+import stirling.software.SPDF.model.api.security.RedactExecuteRequest;
+import stirling.software.SPDF.model.api.security.RedactExecuteRequest.ImageBox;
+import stirling.software.SPDF.model.api.security.RedactExecuteRequest.RedactStyle;
+import stirling.software.SPDF.model.api.security.RedactExecuteRequest.TextRange;
+import stirling.software.SPDF.pdf.parser.PageColumnLayout;
+import stirling.software.SPDF.pdf.parser.PageImageLocator;
+import stirling.software.common.service.CustomPDFDocumentFactory;
+import stirling.software.common.util.ExceptionUtils;
+import stirling.software.common.util.TempFile;
+
+@Service
+@Slf4j
+@RequiredArgsConstructor
+class RedactExecuteService {
+
+    private final CustomPDFDocumentFactory pdfDocumentFactory;
+    private final ManualRedactionService manualRedactionService;
+    private final TextRedactionService textRedactionService;
+
+    TempFile execute(RedactExecuteRequest request) throws IOException {
+        RedactStyle style = request.getStyle() != null ? request.getStyle() : new RedactStyle();
+        List<String> textValues = orEmpty(request.getTextValues());
+        List<String> regexPatterns = orEmpty(request.getRegexPatterns());
+        List<Integer> wipePages = orEmpty(request.getWipePages());
+        List<TextRange> ranges = orEmpty(request.getRanges());
+        List<ImageBox> imageBoxes = orEmpty(request.getImageBoxes());
+
+        boolean hasTargets =
+                !textValues.isEmpty()
+                        || !regexPatterns.isEmpty()
+                        || !wipePages.isEmpty()
+                        || !ranges.isEmpty()
+                        || !imageBoxes.isEmpty()
+                        || request.getRedactImagePages() != null;
+
+        if (!hasTargets) {
+            throw ExceptionUtils.createIllegalArgumentException(
+                    "error.redaction.no.targets", "No redaction targets provided");
+        }
+
+        boolean overlayOnly =
+                RedactExecuteRequest.RedactionStrategy.OVERLAY_ONLY.equals(style.getStrategy());
+        boolean imageFinalize =
+                RedactExecuteRequest.RedactionStrategy.IMAGE_FINALIZE.equals(style.getStrategy());
+        boolean convertToImage = imageFinalize || style.isConvertToImage();
+
+        boolean hasTextOps = !textValues.isEmpty() || !regexPatterns.isEmpty();
+
+        log.info(
+                "[redact/execute] strategy={} textValues={} regexPatterns={} wipePages={} ranges={} imageBoxes={} imagePages={}",
+                style.getStrategy(),
+                textValues.size(),
+                regexPatterns.size(),
+                wipePages.size(),
+                ranges.size(),
+                imageBoxes.size(),
+                request.getRedactImagePages());
+
+        if (request.getFileInput() == null) {
+            throw ExceptionUtils.createFileNullOrEmptyException();
+        }
+
+        PDDocument document = null;
+        try {
+            document = pdfDocumentFactory.load(request.getFileInput());
+
+            // Single-pass text scan: collect all text-based targets so we run the PDF
+            // stripper only once across the entire execute() call rather than once per target.
+            Map<Integer, List<PDFText>> foundTexts =
+                    hasTextOps ? collectTextMatches(document, request) : new HashMap<>();
+
+            int totalMatches = foundTexts.values().stream().mapToInt(List::size).sum();
+            log.info(
+                    "[redact/execute] scan complete: {} text matches across {} pages",
+                    totalMatches,
+                    foundTexts.size());
+
+            // Text removal (content-stream rewriting) — skipped in overlay-only mode.
+            boolean needsOverlayOnly = overlayOnly;
+            if (hasTextOps && !foundTexts.isEmpty() && !overlayOnly) {
+                needsOverlayOnly = applyTextRemoval(document, request);
+            } else if (overlayOnly) {
+                log.info(
+                        "[redact/execute] overlay-only mode requested — skipping content-stream rewriting");
+            }
+
+            // Reload fresh document on fallback so we overlay onto clean content.
+            if (needsOverlayOnly && !foundTexts.isEmpty()) {
+                log.info("[redact/execute] reloading document for clean overlay pass");
+                document.close();
+                document = pdfDocumentFactory.load(request.getFileInput());
+                foundTexts.clear();
+                if (hasTextOps) {
+                    foundTexts.putAll(collectTextMatches(document, request));
+                }
+            }
+
+            // Non-text operations.
+            Map<Integer, PageColumnLayout> layoutCache = new HashMap<>();
+
+            if (!wipePages.isEmpty()) {
+                applyPageWipe(document, wipePages, style);
+            }
+
+            for (TextRange range : ranges) {
+                applyRangeRedaction(document, range, style, layoutCache);
+            }
+
+            for (ImageBox box : imageBoxes) {
+                applyImageBoxRedaction(document, box, style);
+            }
+
+            if (request.getRedactImagePages() != null) {
+                applyAllImagesRedaction(document, request.getRedactImagePages(), style);
+            }
+
+            return manualRedactionService.finalizeRedaction(
+                    document,
+                    foundTexts,
+                    style.getColor(),
+                    style.getPadding(),
+                    convertToImage,
+                    !needsOverlayOnly);
+
+        } catch (Exception e) {
+            log.error("Execute redaction failed: {}", e.getMessage(), e);
+            throw new RuntimeException("Failed to perform PDF redaction: " + e.getMessage(), e);
+        } finally {
+            if (document != null) {
+                try {
+                    document.close();
+                } catch (IOException e) {
+                    log.warn("Failed to close document: {}", e.getMessage());
+                }
+            }
+        }
+    }
+
+    // -----------------------------------------------------------------------
+    // Single-pass text scan (one stripper pass per execute() call)
+    // -----------------------------------------------------------------------
+
+    /**
+     * Runs a single PDF text-stripper pass over all text-based targets and returns the merged hit
+     * map.
+     */
+    private Map<Integer, List<PDFText>> collectTextMatches(
+            PDDocument document, RedactExecuteRequest request) {
+        Map<Integer, List<PDFText>> found = new HashMap<>();
+
+        String[] terms = cleanStrings(request.getTextValues());
+        if (terms.length > 0) {
+            textRedactionService
+                    .findTextToRedact(document, terms, false, false)
+                    .forEach(
+                            (page, hits) ->
+                                    found.computeIfAbsent(page, k -> new ArrayList<>())
+                                            .addAll(hits));
+        }
+
+        String[] patterns = cleanStrings(request.getRegexPatterns());
+        if (patterns.length > 0) {
+            textRedactionService
+                    .findTextToRedact(document, patterns, true, false)
+                    .forEach(
+                            (page, hits) ->
+                                    found.computeIfAbsent(page, k -> new ArrayList<>())
+                                            .addAll(hits));
+        }
+
+        return found;
+    }
+
+    // -----------------------------------------------------------------------
+    // Text removal (content-stream rewriting)
+    // -----------------------------------------------------------------------
+
+    /**
+     * Attempts content-stream text removal for all text/regex targets. Returns {@code true} if the
+     * document fell back to overlay-only mode.
+     */
+    private boolean applyTextRemoval(PDDocument document, RedactExecuteRequest request) {
+        try {
+            boolean fallback = false;
+
+            String[] terms = cleanStrings(request.getTextValues());
+            if (terms.length > 0) {
+                Map<Integer, List<PDFText>> exactFound =
+                        textRedactionService.findTextToRedact(document, terms, false, false);
+                if (!exactFound.isEmpty()) {
+                    fallback |=
+                            textRedactionService.performTextReplacement(
+                                    document, exactFound, terms, false, false);
+                }
+            }
+
+            String[] patterns = cleanStrings(request.getRegexPatterns());
+            if (patterns.length > 0) {
+                Map<Integer, List<PDFText>> regexFound =
+                        textRedactionService.findTextToRedact(document, patterns, true, false);
+                if (!regexFound.isEmpty()) {
+                    fallback |=
+                            textRedactionService.performTextReplacement(
+                                    document, regexFound, patterns, true, false);
+                }
+            }
+
+            if (fallback) {
+                log.warn(
+                        "[redact/execute] font compatibility issue — falling back to overlay-only");
+            } else {
+                log.info("[redact/execute] content-stream text removal applied successfully");
+            }
+            return fallback;
+        } catch (Exception e) {
+            log.warn(
+                    "[redact/execute] text removal failed, falling back to overlay: {}",
+                    e.getMessage());
+            return true;
+        }
+    }
+
+    // -----------------------------------------------------------------------
+    // Per-operation dispatch methods
+    // -----------------------------------------------------------------------
+
+    private void applyPageWipe(PDDocument document, List<Integer> pageNumbers, RedactStyle style)
+            throws IOException {
+        List<Integer> pageIndices = toZeroBasedIndices(pageNumbers);
+        if (pageIndices.isEmpty()) return;
+
+        PDPageTree allPages = document.getDocumentCatalog().getPages();
+        Color pageColor = ManualRedactionService.decodeOrDefault(style.getColor());
+        Collections.sort(pageIndices);
+        log.info("[redact/execute] full-page wipe: {} pages ({})", pageIndices.size(), pageIndices);
+
+        Map<Integer, List<float[]>> pageElementBoxes = new HashMap<>();
+        for (Integer idx : pageIndices) {
+            if (idx >= 0 && idx < allPages.getCount()) {
+                try {
+                    pageElementBoxes.put(
+                            idx,
+                            manualRedactionService.extractPageElementBoxes(
+                                    document, allPages.get(idx), idx));
+                } catch (Exception e) {
+                    log.warn(
+                            "[redact/execute] element extraction failed for page {}: {}",
+                            idx,
+                            e.getMessage());
+                }
+            }
+        }
+
+        for (Integer idx : pageIndices) {
+            if (idx >= 0 && idx < allPages.getCount()) {
+                PDPage page = allPages.get(idx);
+                List<float[]> elementBoxes =
+                        pageElementBoxes.getOrDefault(idx, Collections.emptyList());
+                page.getCOSObject().removeItem(COSName.CONTENTS);
+                page.setResources(new PDResources());
+                try (PDPageContentStream cs = new PDPageContentStream(document, page)) {
+                    cs.setNonStrokingColor(pageColor);
+                    if (elementBoxes.isEmpty()) {
+                        PDRectangle box = page.getBBox();
+                        cs.addRect(0, 0, box.getWidth(), box.getHeight());
+                    } else {
+                        log.info(
+                                "[redact/execute] page {}: drawing {} element boxes",
+                                idx + 1,
+                                elementBoxes.size());
+                        for (float[] r : elementBoxes) {
+                            cs.addRect(r[0], r[1], r[2] - r[0], r[3] - r[1]);
+                        }
+                    }
+                    cs.fill();
+                }
+            }
+        }
+    }
+
+    private void applyRangeRedaction(
+            PDDocument document,
+            TextRange range,
+            RedactStyle style,
+            Map<Integer, PageColumnLayout> layoutCache)
+            throws IOException {
+        String rangeStart = trimOrEmpty(range.startString());
+        String rangeEnd = trimOrEmpty(range.endString());
+        log.info("[redact/execute] range redaction: start='{}' end='{}'", rangeStart, rangeEnd);
+        try {
+            List<PDFText> blocks = collectRangeBlocks(document, rangeStart, rangeEnd, layoutCache);
+            if (!blocks.isEmpty()) {
+                manualRedactionService.redactFoundText(
+                        document,
+                        blocks,
+                        style.getPadding(),
+                        ManualRedactionService.decodeOrDefault(style.getColor()),
+                        false);
+            } else {
+                log.warn(
+                        "[redact/execute] range not found: start='{}' end='{}'",
+                        rangeStart,
+                        rangeEnd);
+            }
+        } catch (Exception e) {
+            log.warn("[redact/execute] range redaction failed: {}", e.getMessage());
+        }
+    }
+
+    private void applyImageBoxRedaction(PDDocument document, ImageBox box, RedactStyle style)
+            throws IOException {
+        List<float[]> boxes =
+                List.of(
+                        new float[] {
+                            (float) box.pageIndex(), box.x1(), box.y1(), box.x2(), box.y2()
+                        });
+        log.info("[redact/execute] image box overlay on page {}", box.pageIndex());
+        Color boxColor = ManualRedactionService.decodeOrDefault(style.getColor());
+        manualRedactionService.redactImageBoxes(document, boxes, boxColor);
+    }
+
+    private void applyAllImagesRedaction(
+            PDDocument document, List<Integer> pageNumbers, RedactStyle style) throws IOException {
+        PDPageTree allPages = document.getDocumentCatalog().getPages();
+        Color imgColor = ManualRedactionService.decodeOrDefault(style.getColor());
+
+        List<Integer> imagePageIndices = toZeroBasedIndices(pageNumbers);
+        if (imagePageIndices.isEmpty()) {
+            imagePageIndices = new ArrayList<>();
+            for (int i = 0; i < allPages.getCount(); i++) {
+                imagePageIndices.add(i);
+            }
+        }
+
+        List<float[]> detectedBoxes = new ArrayList<>();
+        for (int pageIdx : imagePageIndices) {
+            if (pageIdx < 0 || pageIdx >= allPages.getCount()) continue;
+            try {
+                PDPage page = allPages.get(pageIdx);
+                PageImageLocator locator = new PageImageLocator(page, pageIdx);
+                locator.processPage(page);
+                for (PageImageLocator.ImageBox ib : locator.getImageBoxes()) {
+                    detectedBoxes.add(new float[] {pageIdx, ib.x1(), ib.y1(), ib.x2(), ib.y2()});
+                }
+            } catch (Exception e) {
+                log.warn(
+                        "[redact/execute] image detection failed for page {}: {}",
+                        pageIdx + 1,
+                        e.getMessage());
+            }
+        }
+
+        log.info(
+                "[redact/execute] auto image detection: {} images across {} pages",
+                detectedBoxes.size(),
+                imagePageIndices.size());
+
+        if (!detectedBoxes.isEmpty()) {
+            manualRedactionService.redactImageBoxes(document, detectedBoxes, imgColor);
+        }
+    }
+
+    // -----------------------------------------------------------------------
+    // Range collection helpers
+    // -----------------------------------------------------------------------
+
+    /**
+     * Locates {@code startStr} in the document and returns {@link PDFText} blocks for every text
+     * line and image from that point up to (but NOT including) the line where {@code endStr}
+     * begins. If {@code endStr} is blank, redacts from {@code startStr} to the end of the document.
+     *
+     * <p>Multi-column pages follow reading order: down the start column, jump to the top of the
+     * next column, continue to the end anchor. Single-column pages reduce to a plain Y-band check.
+     */
+    List<PDFText> collectRangeBlocks(
+            PDDocument document,
+            String startStr,
+            String endStr,
+            Map<Integer, PageColumnLayout> layoutCache)
+            throws IOException {
+
+        PDPageTree allPages = document.getDocumentCatalog().getPages();
+        int totalPages = allPages.getCount();
+
+        Map<Integer, List<PDFText>> startMatchesByPage = findWithFallbacks(document, startStr);
+        if (startMatchesByPage.isEmpty()) {
+            log.warn("[redact/execute] range start not found: '{}'", startStr);
+            return Collections.emptyList();
+        }
+
+        List<Anchor> starts = toAnchors(document, startMatchesByPage, layoutCache);
+        starts.sort(READING_ORDER);
+        log.info(
+                "[redact/execute] start='{}' matched {} anchor(s): {}",
+                startStr,
+                starts.size(),
+                anchorSummary(starts));
+
+        boolean openEnded = (endStr == null || endStr.isBlank());
+        List<Anchor> ends = new ArrayList<>();
+        if (!openEnded) {
+            Map<Integer, List<PDFText>> endMatchesByPage = findWithFallbacks(document, endStr);
+            if (endMatchesByPage.isEmpty()) {
+                log.warn(
+                        "[redact/execute] range end '{}' not found in document - skipping range"
+                                + " (start='{}')",
+                        endStr,
+                        startStr);
+                return Collections.emptyList();
+            }
+            ends = toAnchors(document, endMatchesByPage, layoutCache);
+            ends.sort(READING_ORDER);
+            log.info(
+                    "[redact/execute] end='{}' matched {} anchor(s): {}",
+                    endStr,
+                    ends.size(),
+                    anchorSummary(ends));
+        }
+
+        List<PDFText> blocks = new ArrayList<>();
+        for (Anchor start : starts) {
+            Anchor end = null;
+            int endPage;
+            if (openEnded) {
+                endPage = totalPages - 1;
+            } else {
+                for (Anchor candidate : ends) {
+                    if (READING_ORDER.compare(candidate, start) > 0) {
+                        end = candidate;
+                        break;
+                    }
+                }
+                if (end == null) {
+                    log.warn(
+                            "[redact/execute] no end anchor after start at (page={}, col={}, y={}) — skipping",
+                            start.page + 1,
+                            start.col,
+                            start.y);
+                    continue;
+                }
+                endPage = end.page;
+            }
+
+            log.info(
+                    "[redact/execute] range pages {}-{}: start='{}' (col {}) end='{}'",
+                    start.page + 1,
+                    endPage + 1,
+                    startStr,
+                    start.col,
+                    openEnded ? "<end of document>" : endStr);
+
+            collectBlocksForRange(document, allPages, start, end, openEnded, blocks, layoutCache);
+        }
+
+        log.info(
+                "[redact/execute] range '{}'→'{}': {} total blocks",
+                startStr,
+                openEnded ? "<end of document>" : endStr,
+                blocks.size());
+        return blocks;
+    }
+
+    /**
+     * Collects all redactable content (text line segments and images) between two anchor positions.
+     *
+     * <p>Line boxes are cached per page number in {@code lineBoxCache} and reused across range
+     * iterations within one execute() call, avoiding redundant {@link AllTextLineExtractor} passes.
+     */
+    private void collectBlocksForRange(
+            PDDocument document,
+            PDPageTree allPages,
+            Anchor start,
+            Anchor end,
+            boolean openEnded,
+            List<PDFText> blocks,
+            Map<Integer, PageColumnLayout> layoutCache)
+            throws IOException {
+
+        int startPage = start.page;
+        int endPage = openEnded ? allPages.getCount() - 1 : end.page;
+        int endCol =
+                openEnded ? layoutFor(document, endPage, layoutCache).columnCount() - 1 : end.col;
+        float startY = start.y;
+        // Use bottom of end anchor so the end anchor line itself is included (inclusive range).
+        float endY = openEnded ? Float.POSITIVE_INFINITY : end.text.getY2();
+
+        // Line-box cache: populated lazily per page, reused across range iterations.
+        // Cannot use computeIfAbsent because AllTextLineExtractor's constructor throws IOException.
+        Map<Integer, List<float[]>> lineBoxCache = new HashMap<>();
+
+        for (int pageIdx = startPage; pageIdx <= endPage; pageIdx++) {
+            PDPage page = allPages.get(pageIdx);
+            float pageHeight = page.getBBox().getHeight();
+            PageColumnLayout layout = layoutFor(document, pageIdx, layoutCache);
+
+            List<float[]> screenLineBoxes = lineBoxCache.get(pageIdx);
+            if (screenLineBoxes == null) {
+                AllTextLineExtractor textExtractor =
+                        new AllTextLineExtractor(pageIdx + 1, pageHeight);
+                textExtractor.getText(document);
+                screenLineBoxes = textExtractor.getScreenLineBoxes();
+                lineBoxCache.put(pageIdx, screenLineBoxes);
+            }
+
+            for (float[] sb : screenLineBoxes) {
+                emitColumnSlices(
+                        pageIdx, layout, sb[0], sb[2], sb[1], sb[3], start.col, startPage, startY,
+                        endCol, endPage, endY, blocks);
+            }
+
+            PageImageLocator imgLocator = new PageImageLocator(page, pageIdx);
+            imgLocator.processPage(page);
+            for (PageImageLocator.ImageBox ib : imgLocator.getImageBoxes()) {
+                // ImageBox coordinates are in PDF user-space (Y up); convert to screen-Y (Y down).
+                float screenY1 = pageHeight - ib.y2();
+                float screenY2 = pageHeight - ib.y1();
+                emitColumnSlices(
+                        pageIdx, layout, ib.x1(), ib.x2(), screenY1, screenY2, start.col, startPage,
+                        startY, endCol, endPage, endY, blocks);
+            }
+        }
+    }
+
+    /** Emits each per-column sub-box accepted by the reading-order predicate. */
+    private static void emitColumnSlices(
+            int pageIdx,
+            PageColumnLayout layout,
+            float x1,
+            float x2,
+            float yTop,
+            float yBottom,
+            int startCol,
+            int startPage,
+            float startY,
+            int endCol,
+            int endPage,
+            float endY,
+            List<PDFText> blocks) {
+        int[] cols = layout.columnsCrossing(x1, x2);
+        if (cols.length == 1) {
+            if (inColumnZone(
+                    pageIdx, cols[0], yTop, yBottom, startPage, startCol, startY, endPage, endCol,
+                    endY)) {
+                blocks.add(new PDFText(pageIdx, x1, yTop, x2, yBottom, ""));
+            }
+            return;
+        }
+        for (int col : cols) {
+            if (!inColumnZone(
+                    pageIdx, col, yTop, yBottom, startPage, startCol, startY, endPage, endCol,
+                    endY)) {
+                return;
+            }
+        }
+        blocks.add(new PDFText(pageIdx, x1, yTop, x2, yBottom, ""));
+    }
+
+    /**
+     * Reading-order predicate: true when (col, yBottom) on page {@code pageIdx} sits between the
+     * start anchor (inclusive) and end anchor (inclusive).
+     */
+    static boolean inColumnZone(
+            int pageIdx,
+            int col,
+            float yTop,
+            float yBottom,
+            int startPage,
+            int startCol,
+            float startY,
+            int endPage,
+            int endCol,
+            float endY) {
+        if (pageIdx > startPage && pageIdx < endPage) return true;
+        if (pageIdx == startPage && pageIdx == endPage) {
+            if (startCol == endCol) {
+                return col == startCol && yBottom >= startY && yBottom <= endY;
+            }
+            if (startCol < endCol) {
+                if (col < startCol || col > endCol) return false;
+                if (col == startCol) return yBottom >= startY;
+                if (col == endCol) return yBottom <= endY;
+                return true;
+            }
+            return col == startCol && yBottom >= startY;
+        }
+        if (pageIdx == startPage) {
+            if (col == startCol) return yBottom >= startY;
+            return col > startCol;
+        }
+        if (pageIdx == endPage) {
+            if (col == endCol) return yBottom <= endY;
+            return col < endCol;
+        }
+        return false;
+    }
+
+    /** Lazily builds and caches the column layout for a single page. */
+    private PageColumnLayout layoutFor(
+            PDDocument document, int pageIdx, Map<Integer, PageColumnLayout> cache)
+            throws IOException {
+        PageColumnLayout cached = cache.get(pageIdx);
+        if (cached != null) return cached;
+        PDPage page = document.getDocumentCatalog().getPages().get(pageIdx);
+        float pageWidth = page.getBBox().getWidth();
+        float pageHeight = page.getBBox().getHeight();
+        AllTextLineExtractor extractor = new AllTextLineExtractor(pageIdx + 1, pageHeight);
+        extractor.getText(document);
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(extractor.getLineBoxes(), pageWidth);
+        if (layout.columnCount() > 1) {
+            float[] g = layout.gutters().get(0);
+            log.info(
+                    "[redact/execute] page {} layout: 2 cols, gutter x=[{}, {}]",
+                    pageIdx + 1,
+                    g[0],
+                    g[1]);
+        } else {
+            log.info("[redact/execute] page {} layout: 1 col (single-column mode)", pageIdx + 1);
+        }
+        cache.put(pageIdx, layout);
+        return layout;
+    }
+
+    private List<Anchor> toAnchors(
+            PDDocument document,
+            Map<Integer, List<PDFText>> matchesByPage,
+            Map<Integer, PageColumnLayout> layoutCache)
+            throws IOException {
+        List<Anchor> out = new ArrayList<>();
+        for (int page : matchesByPage.keySet().stream().sorted().toList()) {
+            PageColumnLayout layout = layoutFor(document, page, layoutCache);
+            for (PDFText hit : matchesByPage.get(page)) {
+                int col = layout.columnOf(hit.getX1(), hit.getX2());
+                out.add(new Anchor(page, col, hit.getY1(), hit));
+            }
+        }
+        return out;
+    }
+
+    /** Lexicographic ordering by (page, column, screenY). */
+    private static final Comparator<Anchor> READING_ORDER =
+            Comparator.comparingInt((Anchor a) -> a.page)
+                    .thenComparingInt(a -> a.col)
+                    .thenComparingDouble(a -> a.y);
+
+    private static String anchorSummary(List<Anchor> anchors) {
+        StringBuilder sb = new StringBuilder();
+        int max = Math.min(anchors.size(), 5);
+        for (int i = 0; i < max; i++) {
+            Anchor a = anchors.get(i);
+            if (i > 0) sb.append(", ");
+            sb.append(String.format("(p=%d,c=%d,y=%.1f)", a.page + 1, a.col, a.y));
+        }
+        if (anchors.size() > max) sb.append(", …");
+        return sb.toString();
+    }
+
+    private record Anchor(int page, int col, float y, PDFText text) {}
+
+    /**
+     * Tries progressively more permissive variants: raw (regex then literal), letter-spacing
+     * collapsed, then a punctuation-tolerant regex over alphanumeric runs.
+     */
+    private Map<Integer, List<PDFText>> findWithFallbacks(PDDocument document, String raw) {
+        String trimmed = raw.trim();
+        String collapsed = collapseLetterSpacing(trimmed);
+        String tolerant = punctuationTolerantRegex(trimmed);
+
+        List<Candidate> candidates = new ArrayList<>();
+        candidates.add(new Candidate(trimmed, true));
+        candidates.add(new Candidate(trimmed, false));
+        if (!collapsed.equals(trimmed)) {
+            candidates.add(new Candidate(collapsed, true));
+            candidates.add(new Candidate(collapsed, false));
+        }
+        if (tolerant != null && !tolerant.equals(trimmed)) {
+            candidates.add(new Candidate(tolerant, true));
+        }
+
+        // If the anchor spans multiple lines (model provided entire paragraph instead of a short
+        // phrase), try just the first non-empty line — it's usually sufficient to locate the
+        // position and avoids mismatches from mid-paragraph text extraction artifacts.
+        if (trimmed.contains("\n")) {
+            String firstLine =
+                    Arrays.stream(trimmed.split("\n"))
+                            .map(String::trim)
+                            .filter(s -> !s.isEmpty())
+                            .findFirst()
+                            .orElse(null);
+            if (firstLine != null && firstLine.length() >= 4) {
+                String firstLineCollapsed = collapseLetterSpacing(firstLine);
+                String firstLineTolerant = punctuationTolerantRegex(firstLine);
+                candidates.add(new Candidate(firstLine, false));
+                if (!firstLineCollapsed.equals(firstLine)) {
+                    candidates.add(new Candidate(firstLineCollapsed, false));
+                }
+                if (firstLineTolerant != null && !firstLineTolerant.equals(firstLine)) {
+                    candidates.add(new Candidate(firstLineTolerant, true));
+                }
+            }
+        }
+
+        for (Candidate c : candidates) {
+            Map<Integer, List<PDFText>> m =
+                    textRedactionService.findTextToRedact(
+                            document, new String[] {c.pattern}, c.useRegex, false);
+            if (!m.isEmpty()) {
+                if (!c.pattern.equals(trimmed)) {
+                    log.info(
+                            "[redact/execute] range boundary matched via fallback: '{}' → '{}'",
+                            trimmed,
+                            c.pattern);
+                }
+                return m;
+            }
+        }
+        return Collections.emptyMap();
+    }
+
+    private record Candidate(String pattern, boolean useRegex) {}
+
+    // -----------------------------------------------------------------------
+    // Static helpers
+    // -----------------------------------------------------------------------
+
+    /**
+     * Joins {@code raw}'s alphanumeric runs with {@code \W*} so anchors match across punctuation
+     * drift. Returns {@code null} when fewer than two tokens exist.
+     */
+    private static String punctuationTolerantRegex(String raw) {
+        List<String> tokens = new ArrayList<>();
+        StringBuilder current = new StringBuilder();
+        for (int i = 0; i < raw.length(); i++) {
+            char ch = raw.charAt(i);
+            if (Character.isLetterOrDigit(ch)) {
+                current.append(ch);
+            } else if (current.length() > 0) {
+                tokens.add(current.toString());
+                current.setLength(0);
+            }
+        }
+        if (current.length() > 0) tokens.add(current.toString());
+        if (tokens.size() < 2) return null;
+        StringBuilder out = new StringBuilder();
+        for (int i = 0; i < tokens.size(); i++) {
+            if (i > 0) out.append("\\W*");
+            out.append(Pattern.quote(tokens.get(i)));
+        }
+        return out.toString();
+    }
+
+    /**
+     * Collapses letter-spaced text produced by position-sorted text extraction.
+     *
+     * <p>When a PDF text stripper runs with {@code setSortByPosition(true)}, letter-spaced headings
+     * come out as {@code "T a b l e o f c o n t e n t s"}. This method converts the spaced form
+     * back to words.
+     */
+    private static String collapseLetterSpacing(String text) {
+        String[] tokens = text.split(" ", -1);
+        StringBuilder result = new StringBuilder();
+        StringBuilder current = new StringBuilder();
+        for (String token : tokens) {
+            if (token.isEmpty()) {
+                if (current.length() > 0) {
+                    if (result.length() > 0) result.append(' ');
+                    result.append(current);
+                    current.setLength(0);
+                }
+            } else if (token.length() == 1) {
+                current.append(token);
+            } else {
+                if (current.length() > 0) {
+                    if (result.length() > 0) result.append(' ');
+                    result.append(current);
+                    current.setLength(0);
+                }
+                if (result.length() > 0) result.append(' ');
+                result.append(token);
+            }
+        }
+        if (current.length() > 0) {
+            if (result.length() > 0) result.append(' ');
+            result.append(current);
+        }
+        return result.toString().trim();
+    }
+
+    private static <T> List<T> orEmpty(List<T> list) {
+        return list != null ? list : List.of();
+    }
+
+    private static String[] cleanStrings(List<String> input) {
+        if (input == null || input.isEmpty()) {
+            return new String[0];
+        }
+        return input.stream()
+                .filter(s -> s != null)
+                .map(String::trim)
+                .filter(s -> !s.isEmpty())
+                .toArray(String[]::new);
+    }
+
+    /**
+     * Converts 1-based page numbers from the request to the 0-based indices used internally.
+     * Out-of-range and non-positive values are silently dropped.
+     */
+    private static List<Integer> toZeroBasedIndices(List<Integer> oneBasedPageNumbers) {
+        if (oneBasedPageNumbers == null || oneBasedPageNumbers.isEmpty()) {
+            return new ArrayList<>();
+        }
+        List<Integer> result = new ArrayList<>();
+        for (Integer page : oneBasedPageNumbers) {
+            if (page != null && page > 0) {
+                result.add(page - 1);
+            }
+        }
+        return result;
+    }
+
+    private static String trimOrEmpty(String s) {
+        return s == null ? "" : s.trim();
+    }
+}

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

@@ -12,6 +12,7 @@ import org.springframework.beans.factory.annotation.Value;
 import org.springframework.core.io.ClassPathResource;
 import org.springframework.core.io.FileSystemResource;
 import org.springframework.core.io.Resource;
+import org.springframework.http.CacheControl;
 import org.springframework.http.MediaType;
 import org.springframework.http.ResponseEntity;
 import org.springframework.stereotype.Controller;
@@ -134,13 +135,22 @@ public class ReactRoutingController {
     public ResponseEntity<String> serveIndexHtml(HttpServletRequest request) {
         try {
             if (indexHtmlExists && cachedIndexHtml != null) {
-                return ResponseEntity.ok().contentType(MediaType.TEXT_HTML).body(cachedIndexHtml);
+                return ResponseEntity.ok()
+                        .cacheControl(CacheControl.noCache().mustRevalidate())
+                        .contentType(MediaType.TEXT_HTML)
+                        .body(cachedIndexHtml);
             }
             // Fallback: process on each request (dev mode or cache failed)
-            return ResponseEntity.ok().contentType(MediaType.TEXT_HTML).body(processIndexHtml());
+            return ResponseEntity.ok()
+                    .cacheControl(CacheControl.noCache().mustRevalidate())
+                    .contentType(MediaType.TEXT_HTML)
+                    .body(processIndexHtml());
         } catch (Exception ex) {
             log.error("Failed to serve index.html, returning fallback", ex);
-            return ResponseEntity.ok().contentType(MediaType.TEXT_HTML).body(buildFallbackHtml());
+            return ResponseEntity.ok()
+                    .cacheControl(CacheControl.noCache().mustRevalidate())
+                    .contentType(MediaType.TEXT_HTML)
+                    .body(buildFallbackHtml());
         }
     }
 
@@ -160,14 +170,19 @@ public class ReactRoutingController {
         return ResponseEntity.ok().contentType(MediaType.TEXT_HTML).body(cachedCallbackHtml);
     }
 
+    // `files` was historically a backend static-asset directory and was therefore
+    // in the exclusion list - removing it lets /files and /files/<folder-uuid>
+    // forward to the SPA index.html, which is what FileManagerView expects.
+    // (Real storage endpoints live under /api/v1/storage/files, already
+    // excluded by the leading `api` token in the same regex.)
     @GetMapping(
-            "/{path:^(?!api|static|robots\\.txt|favicon\\.ico|manifest.*\\.json|pipeline|pdfjs|pdfjs-legacy|pdfium|vendor|fonts|images|files|css|js|assets|locales|modern-logo|classic-logo|Login|og_images|samples)[^\\.]*$}")
+            "/{path:^(?!api|static|robots\\.txt|favicon\\.ico|manifest.*\\.json|pipeline|pdfjs|pdfjs-legacy|pdfium|vendor|fonts|images|css|js|assets|locales|modern-logo|classic-logo|Login|og_images|samples)[^\\.]*$}")
     public ResponseEntity<String> forwardRootPaths(HttpServletRequest request) throws IOException {
         return serveIndexHtml(request);
     }
 
     @GetMapping(
-            "/{path:^(?!api|static|pipeline|pdfjs|pdfjs-legacy|pdfium|vendor|fonts|images|files|css|js|assets|locales|modern-logo|classic-logo|Login|og_images|samples)[^\\.]*}/{subpath:^(?!.*\\.).*$}")
+            "/{path:^(?!api|static|pipeline|pdfjs|pdfjs-legacy|pdfium|vendor|fonts|images|css|js|assets|locales|modern-logo|classic-logo|Login|og_images|samples)[^\\.]*}/{subpath:^(?!.*\\.).*$}")
     public ResponseEntity<String> forwardNestedPaths(HttpServletRequest request)
             throws IOException {
         return serveIndexHtml(request);

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

@@ -22,6 +22,7 @@ import org.springframework.web.bind.annotation.ExceptionHandler;
 import org.springframework.web.bind.annotation.RestControllerAdvice;
 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 jakarta.servlet.http.HttpServletRequest;
@@ -196,12 +197,12 @@ public class GlobalExceptionHandler {
     /**
      * Checks whether the given IOException indicates that the client disconnected before the
      * response could be written (broken pipe, connection reset, etc.). When this happens there is
-     * no point in serialising a {@link ProblemDetail} body because the socket is already closed —
+     * no point in serialising a {@link ProblemDetail} body because the socket is already closed -
      * and attempting to do so may trigger a secondary {@code HttpMessageNotWritableException} if
      * the response Content-Type was already committed as a non-JSON type (e.g. image/png).
      */
     private static boolean isClientDisconnectException(IOException ex) {
-        // Walk the causal chain — Jetty/Tomcat may wrap the low-level SocketException
+        // Walk the causal chain - Jetty/Tomcat may wrap the low-level SocketException
         Throwable current = ex;
         while (current != null) {
             String msg = current.getMessage();
@@ -1040,6 +1041,43 @@ public class GlobalExceptionHandler {
      * @param request the HTTP servlet request
      * @return ProblemDetail with appropriate HTTP status
      */
+    /**
+     * Handle ResponseStatusException explicitly so its embedded HTTP status reaches the client
+     * instead of being swallowed by the {@code RuntimeException} catch-all (which would downgrade
+     * every controller-thrown 400/404/409 to a generic 500). Folder/file storage controllers and
+     * any other code that throws {@code ResponseStatusException} relies on this handler taking
+     * precedence.
+     */
+    @ExceptionHandler(ResponseStatusException.class)
+    public ResponseEntity<ProblemDetail> handleResponseStatusException(
+            ResponseStatusException ex, HttpServletRequest request) {
+        HttpStatus status =
+                HttpStatus.resolve(ex.getStatusCode().value()) != null
+                        ? HttpStatus.valueOf(ex.getStatusCode().value())
+                        : HttpStatus.INTERNAL_SERVER_ERROR;
+        String reason = ex.getReason() != null ? ex.getReason() : status.getReasonPhrase();
+        ProblemDetail problemDetail = createBaseProblemDetail(status, reason, request);
+        problemDetail.setType(URI.create("/errors/" + status.value()));
+        problemDetail.setTitle(status.getReasonPhrase());
+        problemDetail.setProperty("title", status.getReasonPhrase());
+        // 5xx is operator-relevant; 4xx is a normal client-rejection - log at the right level.
+        if (status.is5xxServerError()) {
+            log.error(
+                    "ResponseStatusException {} at {}: {}",
+                    status.value(),
+                    request.getRequestURI(),
+                    reason,
+                    ex);
+        } else {
+            log.debug(
+                    "ResponseStatusException {} at {}: {}",
+                    status.value(),
+                    request.getRequestURI(),
+                    reason);
+        }
+        return ResponseEntity.status(status).contentType(PROBLEM_JSON).body(problemDetail);
+    }
+
     @ExceptionHandler(RuntimeException.class)
     public ResponseEntity<ProblemDetail> handleRuntimeException(
             RuntimeException ex, HttpServletRequest request) {

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

@@ -18,4 +18,11 @@ public class PDFWithPageSize extends PDFFile {
             requiredMode = Schema.RequiredMode.REQUIRED,
             allowableValues = {"A0", "A1", "A2", "A3", "A4", "A5", "A6", "LETTER", "LEGAL", "KEEP"})
     private String pageSize;
+
+    @Schema(
+            description =
+                    "Orientation to apply to the target page size. Ignored when pageSize is KEEP.",
+            defaultValue = "PORTRAIT",
+            allowableValues = {"PORTRAIT", "LANDSCAPE"})
+    private String orientation = "PORTRAIT";
 }