* Security: enforce the automation read boundary on stored smart-filter params
Smart filters store a command id and params inside the document, so a
crafted .pcraft could resolve stored ambient paths (displace mapPath,
color-lookup LUT files) at render time without the authorize gate:
synchronous smart-object rendering has no Session argument, spawned job
threads drop the requesting session's authority, and a trusted UI
render could warm the process-wide stack cache with a stage the gate
never saw.
Thread the requesting session's gate through a thread-local in jobs.rs
(entered around command execution, reinstalled on spawned workers and on
job apply, restored on drop), re-check stored filter commands in
render_in before source resolution and any cache lookup and in
apply_filter_to_surface, and re-authorize layer.smartFilter.setParam
with the complete merged params after merging stored document params —
including edits whose request contains no path and actions.play steps.
Denial is conservative: the document is unchanged and no worker starts.
Regression tests in the automation crate cover rerender, cached-stage
reuse, merged-param edits, action playback, background jobs, and filter
transfer, for both the displace map and color-lookup LUT vectors.
* Review: carry render authority through Session::edit and surface denials
Session::edit is the engine's universal mutation entry (linked-file
refresh, transforms, place); renders reached through it ran with no
authorization scope installed, so stored smart-filter params read
ambiently there even after the execute-path fix. Install the scope in
Session::edit as well, keyed on the session's authorize gate.
apply_filter_to_surface also conflated a denied filter with an unknown
one: authorize_render_command's Result was discarded with .ok()?, so a
denial silently rendered unfiltered pixels and the stage entered the
process-wide stack cache. The signature becomes
Result<Option<Surface>>: Err for denials (callers propagate), None for
ids that aren't pixel filters (pass-through semantics unchanged), and an
unreadable map keeps its deliberate pass-through. apply_smart_filter and
apply_smart_filters propagate instead of cur.clone().
Two tests: a refresh driven outside command execution (Session::edit)
is denied and leaves the document unchanged; a denied render caches no
stage. Nine authorization tests pass, engine suite and strict Clippy
clean, panic hunt green.
* ui_set description: name the layerFilter field main added
The Layers-panel filter dropdown (#5048) added layerFilter to the
control method's UI_SET_FIELDS without extending the ui_set tool's
description, so the drift guard (mcp.rs, ui_set_description_names_every_
accepted_field) fails on Windows CI for any branch built on main. Name
the field with its patch shape: by (kind, name or mode), kinds (a list
of layer kinds), name (matches part of the layer name), mode (a blend
mode name). The 40 MCP tests pass; strict Clippy clean.
* wasm size: compose and algo join cms at opt-level "z" to clear the 24 MiB gate
Main's browser build is 171 KB over the 24 MiB wasm gate in CI
(25,337,343 bytes), so every branch built on current main fails the
web size check. Demote three pixel crates in the wasm-release profile:
- cms: transforms are built once and cached (transform::cached),
so "z" costs little runtime (measured -42 KB).
- compose: the browser app composites per frame on the GPU (crates/gpu,
WebGPU/WebGL2), so the CPU compositor is not the browser's hot path
there (measured -122 KB).
- algo: filters run per pixel on every interactive filter application.
"z" there is the lever of last resort (measured -331 KB), taken
because the gate is enforced in CI; if browser filters feel too slow,
restore 3 here and reclaim the bytes from a feature-gated decoder.
Final size 24,811,949 bytes locally, 354 KB under the gate; the native
profiles are untouched. If a maintainer prefers keeping algo at 3, the
alternative is raising the self-imposed 24 MiB early-warning limit
(Cloudflare's hard cap is 25 MiB) - a policy call in packaging/web/
package.sh, not a code fix.
* Crop tool: Content-Aware fill of the area a rotated or extended crop uncovers (#3823)
image.crop gains contentAware (bool, default false). After a turned crop or one that extends
past the canvas, the area the original image doesn't cover is filled from its surroundings on
the Background (else the active pixel layer), in the crop's one undo step. Each connected piece
(corner, edge strip) gets its own Content-Aware sampling window and samples only covered pixels;
the patch search is Edit > Fill > Content-Aware's (fill_cmds::content_aware_native, shared). A
crop that uncovers nothing is unchanged; one with nothing to sample from fails and rolls back.
Crop options bar: Fill: Background (Default) / Content-Aware Fill (ToolOptions
crop_content_aware, ui.set cropFill). Generative Expand is not offered.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Crop: adapt to Document::background after main removed is_background
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Crop: translate Background (Default) for Hungarian and Turkish
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Fix the merge with main: UI_SET_FIELDS length, Serbian Background (Default)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Merge main: UI_SET_FIELDS counts both new fields
---------
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* feat(commands): find the Elliptical Marquee by 'ellipse' or 'marquee' (#5060)
The Elliptical and Single Row/Column Marquee are options of select.rect
("ellipse":true, "line"), but the command-list filters only looked at
the id and the label ("Rectangular Selection"), so 'ellipse' and
'marquee' found nothing.
Commands can now carry search keywords. command.list / engine.commands
rows include them as "keywords", and the three filters (CLI commands
--filter, the engine.commands RPC and the MCP command_list) share one
matcher over id, label and keywords. select.rect gets the Photoshop
marquee names as keywords.
* Command list replies: document the optional keywords key
The reply-shape test from #3684 pins the exact keys of each row; select.rect
now also carries keywords, so the test and docs/control-protocol.md allow
that one optional string.
* Docs: document the engine.commands reply shape (#3684)
engine.commands, the engine command command.list and the MCP tool
command_list reply with a bare JSON array of command objects, which the
docs never said. Document the shape (keys, types, an example element,
how filter/enabled_only narrow it) in control-protocol.md and mcp.md,
state it in the MCP tool description, and pin it with an rpc test so
the docs cannot drift. The shape itself is unchanged, so existing
agents and scripts keep working.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Command list docs/test: include main's disabledReason and needsDocument keys
Main added `disabledReason` and `needsDocument` to every command list row
(#3664), so the documented shape and the pinning test now list all eight keys.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Draw the active layer's storage tiles over the canvas: a faint
TILE_SIZE grid over the document, every stored tile filled and
labelled tx,ty, tiles with identical pixels in one colour per group
and unique tiles in blue. It shows how sparse a layer is, which tiles
copy-on-write shares, and which ones .pcraft saves once.
It has no menu item and isn't a View > Show item. The control channel
and MCP turn it on with ui.set {"tileGrid": true}; ui.inspect reports
it as view.tile_grid, and View > Extras doesn't hide it.
Grouping hashes each distinct tile allocation once, in parallel (or
serially while jobs hold the pool, #1554), cached per document
revision and active layer. A 36 MP layer of all-different tiles groups
in 3.9 ms (8-bit) and 7.5 ms (16-bit) on 24 cores in release; the
ignored tile_grouping_36mp test reproduces this.
Also documents tileGrid in docs/control-protocol.md and
docs/development.md, and adds my contributors/people.toml entry.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Read SMPTE 268M DPX in both byte orders (SDPX, XPDS): the first image
element, descriptor 6 (luma), 50 (RGB) or 51 (RGBA), at 8, 10, 12 or 16
bits, with 10-bit packing 0, 1 (method A) or 2 (method B) and 12-bit
packing 0 or 1. 10- and 12-bit code values open as U16
round(c * 65535 / (2^b - 1)) and come back exactly when written at their
depth. No colour transform: the transfer characteristic and colorimetric
specification are reported as import notes; extra image elements are
reported; the orientation is applied on import.
Write big-endian, one unencoded element, at 8, 10, 12 or 16 bits
(EncodeOptions::dpx_bits; `dpxBits` on the save commands, MCP and RPC;
`--dpx-bits` in the CLI), following the document depth by default. Gray
with alpha is written as luma. Creator, project, copyright, creation
time, transfer, colorimetric, frame position, frame rate and time code
travel as DPX:* text metadata and are written back.
Unsupported descriptors, signed or run-length-encoded data and other bit
depths return CodecError::Unsupported; malformed and truncated files
return errors.
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
engine.commands, command.list and the MCP command_list tool now give each
row a disabledReason (null when enabled, else the failed precondition) and
needsDocument (true when the command is off because no document is open),
so an agent can tell "open a document first" from "not runnable". enabled
keeps its meaning. Both listings share one row builder in the engine.
Fixes#3664
Co-authored-by: Fabio Lussi <198260595+itrycodinganditsbad@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Agents read the tool description, not docs/mcp.md, and the command_batch
description only said it stops at the first error. It now also says the
steps run in order and that earlier successful steps stay applied: there
is no rollback. docs/mcp.md gives the stop_on_error default and a recovery
recipe: inspect with doc_inspect, undo with edit.undo once per history
entry (undo does not reverse files written to disk), or experiment on a
copy made with image.duplicate or history.newDocument. A test checks the
listed description.
Claude-Session: https://claude.ai/code/session_01BSB7fMQUccqFQekksDGWpo
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* automation: doc.place reads a file through the read root (#2810)
file.placeEmbedded takes an ambient path, so automation refuses it and an
agent could not place a file into the open document at all. doc.place
resolves the path beneath the read root the way doc.open does and hands the
bytes it already read to the public file_cmds::place_bytes, so the existing
cap-std rules stay the only path rules: no new scope logic, and the ambient
command keeps its refusal. Trusted-local callers still run the command.
* automation: export layers, data sets and CSVs through the roots (#3216)
file.export.layersToFiles and file.export.dataSetsAsFiles take an ambient
dir, file.import.variableDataSets an ambient path, so all three are refused
in automation even for files inside the granted roots, and an agent could
not export a single file or a data set at all.
Both halves already existed: the engine writes bytes it has just encoded
(save_doc is encode + write_file, and the data-set exporter encodes with
photocraft_io::export before writing), and the automation side already
holds a write capability — AuthorizedWorkspace::export_sequence creates the
directory, refuses to overwrite, accepts only single-component names and
removes every file it created when a step fails, which is what Render Video
uses. So each command body moves behind the same sink callback
render_video_with takes: the command keeps writing the bytes it is handed
to its dir, the new methods write the very same bytes through the root, and
the two implementations cannot drift.
---------
Co-authored-by: switch87 <8284204+switch87@users.noreply.github.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
file.placeEmbedded takes an ambient path, so automation refuses it and an
agent could not place a file into the open document at all. doc.place
resolves the path beneath the read root the way doc.open does and hands the
bytes it already read to the public file_cmds::place_bytes, so the existing
cap-std rules stay the only path rules: no new scope logic, and the ambient
command keeps its refusal. Trusted-local callers still run the command.
Co-authored-by: switch87 <8284204+switch87@users.noreply.github.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Content-Aware Move: Transform On Drop (#3071)
The Content-Aware Move tool gains Photoshop's Transform On Drop checkbox. With it on, the drop
puts a transform box on the moved patch (the Free Transform box's handles, limited to move,
scale and rotate); Enter, the check button or a double-click computes the content-aware blend
with the transformed patch, Esc or the cancel button drops the whole move. The document is
untouched until the commit, so the move stays one History step.
Engine: paint.contentAwareMove takes scaleX / scaleY (%, 1..10000, negative flips) and rotate
(degrees, clockwise) about the selection's centre before the offset, validated (Err, not a
panic, for non-finite, out-of-range or off-canvas placements). The content and the selection's
shape are resampled through the inverse map on float samples (any depth); the selection is
transformed with the content. A plain move keeps the old, exact code path.
UI: cam_transform.rs holds the box (UiState.cam_transform, view state); the option is remembered
in ToolOptions and settable with ui.set camTransformOnDrop; ui.inspect reports camTransform.
Scorecard TOOL-213-4 stays partial: no live preview of the blended result yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Tests: keep Transform On Drop cases in the Content-Aware Move test, not Patch (#3071)
paint.patch takes no scaleX/scaleY/rotate, so those cases (which only
paint.contentAwareMove rejects) failed patch_fails_gracefully. They are
already covered by content_aware_move_fails_gracefully.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* MCP bridge: reconnect before idle sockets die, retry reads, actionable timeouts (#2487)
The app closes idle control connections after 30s (IO_TIMEOUT) while the
bridge kept one persistent connection, so the first call after a pause
failed with "operation may have completed". The bridge now drops and
reconnects before sending once the connection has been idle 25s, so a
request is never written to a socket the app may have dropped.
The default call timeout drops from 60s to 55s, strictly below the app's
60s reply deadline, so the bridge's own error — naming the method and the
likely causes — is the one callers see instead of a race between the
app's deadline reply and the caller's own cutoff.
Idempotent read methods (ui.inspect and friends) get one automatic retry
on a transport failure; edits keep #1007 semantics and are never resent.
A timed-out call is never retried (it would only double the wait).
* MCP bridge: caller-vouched read retries, per-failure wording, docs contract (#2487 review)
The retry decision moves from a hand-maintained method list — most of whose
entries were engine command ids that never travel as control-protocol
methods, so they could never match — to the call sites that know a call is
a read: call never resends (#1007), call_read retries once on a transport
failure, and the MCP server opts its pure-read tools in. Failures are
worded by what they prove: a connection that could never be established is
"the request was not sent", not "may have completed"; a read timeout says
a read may be retried safely instead of demanding a state inspection. The
55s client cap, the read-retry exception, and the idle reconnect are now
documented in the control protocol and the security book.
Two Studio-layout themes for KDE Plasma desktops: original colour values
chosen to sit close to Breeze's cool blue-grey surfaces and blue accent
(and its 5 px radius) so the app blends into a Plasma desktop, without
copying Breeze's colour schemes. Text readability comes first.
- theme.rs: ThemeKind::BreezeLight / BreezeDark. Every text colour (text,
secondary, faint, accent, danger, warning) holds 4.5:1 on every surface
and on hovered, pressed and selected states; the primary button is a
deeper blue that takes white text and stays above 4.5:1 while it
darkens on hover and press. Breeze's own inactive text and link blue
are only about 2.8:1 on its light surfaces, so ours are darker. The
close button goes red on hover.
- prefs: Theme/LightTheme::BreezeLight ("breezeLight") and
Theme/DarkTheme::BreezeDark ("breezeDark"): Preferences > Interface,
`prefs.set`.
- Window > Theme > Breeze Light/Dark Theme (`window.theme.breezeLight`,
`window.theme.breezeDark`) and `ui.set {"theme": ...}` (aliases:
breeze, kde, plasma for the light one; ...dark for the dark one).
- i18n: the four new labels in all 16 catalogs.
- docs/ui-design.md and ATTRIBUTION.md (original work, credits Breeze as
the inspiration; Breeze's schemes are LGPL-2.0-or-later and none of
their values, files or code are copied).
- Tests: names/aliases, surface order and blue-grey character, the
contrast matrix for both themes, and the preference/mode selection.
* Crop tool: Photoshop's default mode turns the image behind an upright box; Use Classic Mode (P), Auto Center Preview (#1919)
The Crop tool's gear menu gets Use Classic Mode (P toggles it while the tool shows a box; off by
default, as in Photoshop) and Auto Center Preview (on by default). In the default mode the box
stays upright on screen: a drag outside it turns the image about the box's centre (Shift 15 deg
steps, angle readout), a drag inside moves the image under the box, the arrow keys nudge the
image, Straighten turns the image so the line is level, and with Auto Center Preview a drawn or
resized box is re-centred in the canvas. Classic Mode is the previous behaviour.
The crop model and the commit are shared: crop_rect + crop_angle and one image.crop {angle}.
crop_mode.rs only drives the view: while a frame is edited the view's rotation is -crop_angle,
pivoting so the box (or the pressed point) stays put, and the user's view (a Rotate View turn,
its centre) is put back on commit, cancel, leaving the tool or switching to Classic Mode. The
untouched default frame leaves an R turn alone. The options live in ToolOptions.crop_shield
(classic_mode, auto_center_preview; ui.set cropShield). ui.pointer reads a default-mode drag's
points in the view at its press, so an up where the last move was is idempotent.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Crop default mode: match Photoshop (view upright on tool pick, live Auto Center, image-turn readout, gear menu order) (#1919)
Checked by hand against Photoshop in its default (non-classic) mode:
- Picking the Crop tool turns a Rotate View (R) turn upright at once (about the view centre),
and the view stays upright after Enter or Esc: the save/restore of the user's view is gone.
Classic Mode still leaves the view alone (not checked against Photoshop).
- The angle beside the pointer while the image turns is the image's turn, clockwise positive.
- Auto Center Preview keeps a box being resized centred live: the handle follows the pointer,
the box grows on both sides, and on the image the opposite edge (with Alt the centre) stays
put. Computed from the press, so the view following the frame can't feed back into the drag.
- Gear menu order as Photoshop's: Use Classic Mode (P), Show Cropped Area (H), Auto Center
Preview, no separator; Classic Mode greys out the other two. Show Cropped Area then acts as on
and H does nothing (the stored choice is kept). German: Classic-Modus verwenden, Automatische
zentrierte Vorschau.
- Crop preview past the canvas in a turned view: each band is drawn as its turned quad (the
screen bounding box stretched band textures and spread the checkerboard). The white corners in
the earlier offscreen snapshot were egui_kittest clamping the repeating checker texture; the
real app shows the checkerboard in both modes.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* docs: regenerate scorecard for the crop default-mode row (#1919)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Crop default mode: the canvas keeps the view the Crop tool moves; a move starts from a centred box (#1919)
Tried by hand in the app: with Auto Center Preview on, a drag inside the box made the box wander
instead of moving the image, and a handle drag didn't keep the box's centre in place (after a
Space pan, too). Root cause: canvas_view works on a copy of the view for the frame and writes it
back at the end, so every camera change the Crop tool made during the frame (the image moving
with the pointer, the live-centred resize, Auto Center on release, the pivot while turning) was
undone, while the frame itself moved in the document. Tests driving tool_event never saw it.
- canvas_view notes the camera before the frame's input and, before writing its copy back,
keeps a centre or rotation a tool changed meanwhile.
- With Auto Center Preview a move inside an off-centre box (after a pan) centres the box first,
then grabs the image point now under the pointer.
- New tests/crop_default_mode.rs drives the full app UI with egui pointer and key events over
several frames (offscreen wgpu): the box stays put while the image follows the pointer, and a
resize after a Space pan keeps the box's centre fixed while the handle follows the pointer. Both
failed before the fix.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Crop default mode: a moved or resized box stays where it is on screen (#1919)
Checked in Photoshop: after a Space (Hand) pan the box isn't snapped back to the middle of
the canvas. A drag inside moves only the image, the box staying wherever it is; a resize keeps the
box's centre where it was at the press (also after a pan), and its release no longer pans the box
to the middle. Auto Center Preview still centres a newly drawn box on release and X still
re-centres. Classic Mode is unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Crop tool: Classic Mode turns a Rotate View turn upright too (#1919)
Checked in Photoshop: with Classic Mode on and the view turned with Rotate View (R), picking
the Crop tool also turns the view upright, and the crop frame stays on the same image area. So the
Crop tool now drives the camera in both modes: default mode keeps it at -crop_angle (the box
upright), Classic Mode at 0 with the frame drawn with its own turn. Picking the tool pivots about
the view centre; commit and cancel leave the view upright in both modes, and P switching modes
never brings the R turn back. Auto Center Preview stays default-mode only.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Regenerate the scorecard; update the crop test's header
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Crop: snap a new box's corner inside the untouched default frame in both modes (#2549 with #2545)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Expose native shape stroke controls with staged previews
* Document shape stroke defaults in MCP control schema
* Regenerate docs/scorecard.md
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: optimalbytee <optimalbytee@users.noreply.github.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
doc_open, doc_save and the other automation file paths only took
forward-slash relative paths, so an agent passing C:/Users/me/a.png with
a C:/ root was refused. An absolute path is now compared with the
root's absolute path as text: if it lies beneath the root, the root's
components are stripped and the rest goes through the same checks as a
relative path and is opened through the same directory capability.
Nothing is resolved on the filesystem, so traversal, prefixes, device
names and link escapes are refused as before; absolute paths outside
the root, or with no root granted, are still refused.
Fixes#2176.
* Brush preset chip: draw the tip preview in the icon colour on light themes
The options-bar chip always painted the round tip white, which nearly vanished on Studio
Light's chrome (246 gray) and disappears entirely on a white header bar. Dark themes keep
Photoshop's white tip; light themes use their icon token.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Add Adwaita and Adwaita Dark themes
Two opt-in themes that put the Studio layout in GNOME's libadwaita palette, so PhotoCraft
sits with the rest of a GNOME desktop: header bar, sidebar, card and view colours, the
GNOME blue accent (#3584e4), destructive and warning colours, 6 / 9 / 12 px radii, and a
neutral (not red) close button as in GNOME's title bars. The default theme is unchanged.
Wired through ThemeKind, the interface.theme preference (adwaita, adwaitaDark), Window ›
Theme and the control channel, with menu and preference labels in every complete-menus
language ("Adwaita" kept as the proper name). A test holds both themes to WCAG contrast:
7:1 for text and 4.5:1 for dim text on every surface that carries labels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* i18n: translate the Adwaita theme names in Dutch and Ukrainian
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
The Crop options bar gets Photoshop's gear menu (Set additional Crop options): Show Cropped Area
(H toggles it while the Crop tool shows a box; off hides everything outside the box behind the
canvas colour) and Enable Crop Shield with Color (Match Canvas, the pasteboard colour, or Custom…,
which opens the Color Picker), Opacity (default 75 %) and Auto Adjust Opacity (half the opacity
while the box is dragged). The shield used to be a fixed black at about 51 %. The options persist in
ToolOptions.crop_shield (older settings get Photoshop's defaults; a stray opacity is clamped) and
ui.set takes cropShield. Classic Mode and Auto Center Preview are left out: PhotoCraft always crops
in Classic Mode.
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The Crop options bar's overlay button opens Photoshop's overlay menu: six overlays, Auto/Always/Never
show modes, and O / Shift+O to cycle the overlay and its orientation. The default is Rule of Thirds with
Auto Show Overlay, as in Photoshop. ui.set takes cropOverlay, cropOverlayShow and cropOverlayOrientation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every tool's input schema is rewritten once at startup: $refs inlined (use-site
descriptions win), $defs dropped, boolean schemas turned into {}, with a depth cap
for recursive definitions. ui_pointer events get a concrete schema. Strict providers
no longer reject the whole tool list. Fixes#1782.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The MCP server follows the crafts' shared conventions (FilmCraft #28): core tools
incl. render_preview, titles and annotation hints, strict argument keys (-32602),
live photocraft://document and photocraft://commands resources, and -32700
recovery from malformed JSON lines. Headless Render Video writes through the
granted write root, reports progress, stays ping-responsive and can be cancelled,
removing only the files it created. docs/mcp.md describes it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Eyedropper options bar now has Photoshop's Sample Size (Point Sample,
3 by 3 ... 101 by 101 Average), Sample (Current Layer, Current & Below,
All Layers, All Layers No Adjustments, Current & Below No Adjustments)
and Show Sampling Ring.
- engine: new `document.sampleColor {x, y, size, sampleLayer}`
(sample_cmds.rs). Averages the square centred on the pixel, clipped to
the canvas, alpha-weighted, in the document's colour model (Gray, Lab;
CMYK documents use their composite RGB, since the profile round trip
is not exact) and rounded to the document's bit depth.
- ui: eyedropper_ui.rs options bar; state.rs ToolOptions
eyedropper_size / eyedropper_sample / eyedropper_ring (serde). The
size is shared with the painting tools' Alt-click eyedropper, the
Curves / Color Picker eyedroppers and the Info panel readout.
- control: ui.set eyedropperSampleSize / eyedropperSample /
eyedropperRing; documented in control-protocol.md and the MCP schema.
- i18n: new strings in all 14 catalogs (plus the missing Russian hint
the tl! scan had not reached).
- scorecard: TOOL-1649-1.
- #1599 tests that ui_set's description names every field ui.set accepts; #1611 adds
`rotation` to ui.set, so the description now names it.
- #1611's el.tsv row for "Pick screen color" duplicates the one #1618 added on main.
- docs/scorecard.md regenerated with cargo xtask scorecard after merging both.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UHgFEHEsefoQoBParH47JA
The ui_set tool's `fields` parameter description listed 18 of the 21
fields the desktop `ui.set` control method accepts, omitting colorPanel,
gradientBlendMode and gradientClassic, and then ended with "Other fields
are an error". An agent that copies the list out of tools/list is told
three working fields are invalid.
The description now names all three, in the order UI_SET_FIELDS lists
them, with what each takes: colorPanel ({background: bool}) and
gradientClassic (bool) and gradientBlendMode (a blend mode name).
The new test reads UI_SET_FIELDS from the desktop shell's source and
asserts the schema names every entry, so the list cannot drift again --
which is how this omission arrived. It fails on colorPanel before this
change, and also fails if a later field is added to UI_SET_FIELDS without
updating the description. The parse is anchored on the declaration and
checked against its declared array length, because the doc comment
mentions UI_SET_FIELDS first and scraping that reads unrelated names.
Co-authored-by: Space Bunny Free <chasbigham@gmail.com>
* Right-click Brush Preset picker: pick presets like Photoshop (#615, #1031)
One picker for the canvas right-click and the options-bar brush chip, drawn
once a frame and kept in UiState (brush_picker, brush_picker_list). Its
preset list is the Brushes panel's (brushes_tab::preset_list): a click picks
and keeps it open, a double-click picks and closes it, drag reorders and
right-click renames or deletes. A gear menu adds New Brush Preset… (named in
the rename bar), Rename/Delete Brush, list or grid view and Import Brushes….
Hardness shows for round tips only; Enter applies a typed size before it
closes the picker. ui.set gains brushPicker and brushPickerView.
Also fixes the Brushes panel's Rename Brush…/Rename Group… context items,
which never opened the rename bar.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Brush Preset picker: Polish translations
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UHgFEHEsefoQoBParH47JA
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The bridge branch forwarded only the path to the running app's
app.save, silently dropping quality, format, tiffLayers and index (the
headless branch honours all of them): a client asking for a
quality-50 JPEG got an ok reply and whatever the app's current
settings produce. Those params now fail fast with the reason - naming
headless mode - before anything is written.
* Lossy WebP: a pure-Rust VP8 encoder with quality and lossless controls (#648)
crates/codecs gains codecs/vp8: a clean-room VP8 key-frame encoder from RFC 6386
(boolean coder, 16x16 and 4x4 intra prediction with per-macroblock mode decision,
forward DCT/WHT paired with the bit-exact inverse transforms, dc/ac quantiser tables,
token coding with per-frame probability updates, skipped macroblocks). webp.rs wraps
the frame in the RIFF container (bare VP8, or VP8X with ALPH alpha and ICCP/EXIF/XMP).
EncodeOptions: webp_lossless now picks the encoder instead of erroring, webp_quality
(0-100) sets the quantiser; a lossy WebP reports a LossyCompression fidelity warning.
Export As: the WebP entry gets Quality and a Lossless switch (new tl! string in all
catalogs). CLI --quality, MCP/RPC doc_save quality and the engine's JPEG-style
quality turn a WebP lossy; without a quality WebP stays lossless as before.
Tests: tests/webp_lossy.rs decodes every file with image-webp (PSNR by quality,
sizes fall with quality and are >= 4x below lossless on noisy photos, flat frames
under 1.2 KB, alpha exact, metadata round trip, odd/tiny sizes, 16-bit input, the
16383 px limit as an error, text edges), plus unit tests for the coder, transforms,
prediction and header.
Co-Authored-By: Claude Mythos 5.1 <noreply@anthropic.com>
* i18n(it): Lossless (Export As WebP switch) in the Italian catalog that landed on main
Co-Authored-By: Claude Mythos 5.1 <noreply@anthropic.com>
---------
Co-authored-by: leon <claude.udder842@passmail.net>
Co-authored-by: Claude Mythos 5.1 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Maintainer decision on #672: the app's Save As keeps TIFF layers as Photoshop does (it sets
ExportOptions.tiff_layers from its Layers option), but scripted and agent saves stay flat by
default so conversions keep predictable sizes. #672 merged with the default on, so every
headless TIFF save became layered.
- photocraft-io: ExportOptions::default() has tiff_layers: false. The desktop and web Save As
already set it explicitly from their settings, so the app is unchanged.
- photocraft-cli: --tiff-layers on convert, run and batch (listed in the usage text).
- engine: file.saveACopy, Batch, Image Processor, the export-to-files commands and the lens
batch take "tiffLayers": bool=false, through one SaveOpts { quality, tiff_layers }.
- automation: serve doc.save and MCP doc_save take tiffLayers (default false).
- Tests: default ExportOptions write a flat TIFF; the CLI writes flat without the flag and
layered with it (run and convert); #672's layered-TIFF tests ask for layers explicitly.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Color Picker: sample colours from the image with a pipette
With the Color Picker open, the pointer over the image turns into a
pipette, as in Photoshop. A click or drag there samples the composite
colour into the picker's new colour, and OK applies it. This works
whatever tool is selected; Space and middle-drag still pan. Over the
dialog the pointer is the normal arrow.
Agents get the same: `ui.pointer` (MCP `ui_pointer`) samples into an
open picker instead of driving the tool.
* Color Picker tests: rename the eyedropper test's colour helper after the rebase
#485 added a `color` test helper in the same module.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Command params that are not a JSON object are an error (#506)
`Session::dispatch` now rejects array, string, number and boolean params
with "invalid parameters for `<id>`: params must be a JSON object" instead
of running the command with every default; `null` still means no params.
The one check covers the control channel, `serve`, MCP (headless and
bridge) and the CLI, and batch steps fail per step. `photocraft-cli run
--new` prefixes its error with `--new`.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Headless requests apply finished background jobs first (#503)
A `wait: false` job was applied to its document only by a later command
or `jobs.list`, so `doc.save`, export, `doc.inspect`, `doc.render` and
`session.list` (and their MCP tools) still read and wrote the document
without it. `Headless::sync_jobs` now runs at the top of every JSON-lines
request and before every headless MCP tool call.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* `serve` stdio keeps the session after a rejected request line (#505)
An over-long request line (or one that is not UTF-8) ended the stdio
session, losing every open document. It now gets one error reply, the
rest of that line is skipped without being dispatched, and the next
request is served. `read_bounded_line` reads bytes, so a cut inside a
multi-byte character is reported as too long instead of an I/O error.
TCP connections still close after an oversized line.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* MCP `command_batch` keeps its step results when the reply budget stops it (#504)
The batch budget charged each step's plain JSON size, but MCP sends the
batch reply as escaped text content, so a batch the budget had already
stopped could still exceed the tool-result ceiling and come back as a
bare "response exceeds" error. MCP (headless and bridge) now uses
`BatchReplyBudget::escaped`, which charges the escaped size, through the
new `Headless::batch_with_budget`; `serve` keeps the plain budget.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs: the 1 MiB request-line limit does not apply to MCP stdio (#529)
The book stated the 1 MiB request-line limit without naming a transport.
It applies to desktop and headless TCP and to the headless JSON-lines
stdio server; MCP over stdio reads its own lines and only its tool
results are size-checked. Say so in the automation security pages and in
the control protocol's transport limits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* ui.inspect descriptions point to ui.menu.list for the menu tree (#510)
The control-protocol doc, the control.rs method list and the MCP
`ui_inspect` description said `ui.inspect` returns the menu tree; it
never did. They now say the menu tree comes from `ui.menu.list` (over
MCP, through `control_call`).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* ui.menu.invoke and ui.dialog.confirm wait for the job they start (#511)
Since filters became background jobs, invoking a no-dialog filter from
the menu or confirming a filter dialog over the control channel replied
`{job, pending: true}` before the filter was applied, unlike v0.2.0 and
unlike `engine.execute`. All three now share one helper that runs the
command with script events off and, when it started a job, defers the
reply until the job's result lands; `"wait": false` keeps the immediate
`{job, pending}` reply. The now-unused `run_automation` is gone.
The new test drives Blur More through `ui.menu.invoke` and Gaussian Blur
through its dialog's OK, with and without `wait` (it failed before).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* ui.pointer right-click opens its menu at the pressed point (#512)
A `ui.pointer` right-click listed the layers at the requested document
point but drew the layer menu at the canvas centre, and the Brush Preset
picker opened there too. Both now open at the point's screen position,
where a mouse right-click on it opens them. `ViewXform::active` gives the
active canvas's document-to-screen mapping as last laid out (the
construction several tests repeat); the centre stays the fallback when
the point can't be mapped.
The ui.pointer doc now lists `button`, both menus, and that top-level
modifier flags are ignored when `modifiers` is given.
The new test checks two points against the mouse path and the picker
position (it failed before).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* MCP ui_pointer forwards button and rejects unknown arguments (#514)
Bridge clients couldn't right-click through `ui_pointer`: it forwarded
only `events` and `modifiers` and silently dropped anything else
(`button`, a typo, a top-level `space`). It now forwards an optional
`button`, and `PointerParams` denies unknown fields, so the schema says
`additionalProperties: false` and a stray argument is an error.
The new bridge test checks the forwarded params, the rejection of
`buton` and top-level `space`, and the tool schema (it failed before).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
filter.blur.gaussian isn't registered, so the documented CLI run and the
agent loop failed with "unknown command". Pin the schema examples with a test.
MCP `doc_save {}` and `serve` `doc.save {}` fell back to the document's
own path for any format, so an agent that saved its work replaced the
input PNG or JPEG with a flattened, re-encoded copy, and a `format`
override wrote PNG bytes under a `.psd` name. The desktop's File › Save
already writes back only to layered files.
The rule now lives in one place, `engine::file_cmds::saves_in_place`
(plus `extension`), shared by File › Save, the control channel's
`app.save` and the headless save. Without `path`, a save writes back
only to a PSD, PSB or .pcraft file in its own format; anything else is
an error naming `path` and the file is left unchanged. An explicit
`path`, even the opened file's own, still writes as before. The MCP
`doc_save` description and `path` doc, and docs/control-protocol.md,
state the rule.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
`command_batch` steps reuse `command_run`'s params, so its schema
advertised `wait`, but the tool mapped each step to `{command, params}`
and ran it to completion: `"wait": false` blocked on the filter and
started no job. Steps now carry `wait` (default true): headless batches
start the command through `command_start`, and bridge mode forwards it
to `engine.execute`, which already honours it. `serve`'s `batch` reads a
step's `wait` the same way. The schema doc and docs/control-protocol.md
describe the per-step `wait`.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
`ui.set` skipped keys it doesn't read and replied with success, so the
`dark` field advertised by the MCP `ui_set` tool (and any typo) changed
nothing while reporting ok. It now checks every key against
`UI_SET_FIELDS` before applying any of them and names the accepted
fields in the error. The MCP tool description, its `fields` doc, the
control.rs method list and docs/control-protocol.md list the real fields.
An unknown theme listed only studio, studioLight and classic; the error
now comes from `ThemeKind::ALL` through the new `ThemeKind::id`, so it
names pro and proMedium too, and each listed name selects its theme.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Test: retry the GPU startup sentinel check while a forked child briefly shares the lock
monitor_profile tests spawn osascript in the same binary; between fork and exec the child holds
a copy of the marker file descriptor, so flock reports WouldBlock and begin() returns Busy.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Background jobs: engine job API, progress and cancel (part of #210) (#289)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
