* claim: #7181
* MCP save: a folder the agent opened or saved a design in stays in scope (#7181)
The agent's default folders held the open design's folder only while that
design stayed open there, so after a new design, an opened STEP file or a
save-as elsewhere a save beside the earlier design was refused, while the
same save through the control port worked.
The MCP server now keeps, for the rest of its run, the folder of every
design file a call opened or saved (open, save, and doc.open, file.save,
file.save_as through execute or batch, path given in full) and, headless,
of every design open in its session. Such a call only succeeded because the
folder was already in scope, so naming a path adds nothing. The folders get
the open design folder's terms (inside home or temp, not hidden, hidden
entries unreachable), at most 16. Bridged, they go to the app as
design_folders, which it judges the same way.
The save tool description and docs/mcp.md say how to add a folder with
--allow-path.
* MCP scope: a design's folder is the folder its file resolves to (#7181)
A link in another folder to a design inside the agent's folders passes the
scope check, which follows it, but its own folder was what got remembered
(and, for the open design, what counted while it was open). Take the folder
of the file as the check resolves it.
* Control test: a hidden file the export format accepts (#7181)
---------
Co-authored-by: MateuszSychXpl <151755914+MateuszSychXpl@users.noreply.github.com>
* claim: #7635
* Keep the drag cursor on the manifold of an ill-conditioned sketch (#7635)
A construction line collapsed to zero length makes its angle row's entries
on that line ~1e7 while the same row's entries on the measured line are ~1.
Dragging an endpoint of the midpoint-pinned line then failed two ways: the
null-space rank analysis, which scaled rows only, hid the small (physically
meaningful) entries below its pivot tolerance and let the measured line
rotate freely; and the cursor's refinement, which demanded a TOL-tight
correction the ill-conditioned system cannot deliver, gave up entirely and
let the drag leave an unsolvable step placed - breaking the perpendicular
and angle constraints.
Equilibrate the null-space Jacobian on both sides (columns by their largest
entry, then rows) so one row's entries share a scale, and let the cursor
keep working from a pose the solver itself calls solved (residual below
ACCEPT) when refinement stalls above TOL, accepting corrected trials at the
same tolerance the final validation uses.
---------
Co-authored-by: Dahak Builds <dahakbuilds@users.noreply.github.com>
* agent chat: OpenCode Go preset with per-model endpoints
One preset (https://opencode.ai/zen/go/v1) covers every Go model: each
model is routed to its endpoint family (chat completions, responses,
messages, Gemini), always with the x-opencode-session header Go
requires. Unknown models fall through protocols when Go names the
mismatch; the conversation and tools translate per protocol.
* agent chat: keep Go reasoning metadata, route all listed models (#5656)
* fix agent chat Go endpoints to documented inference routes
* style: collapse long assert in Go routing test (cargo fmt)
* refactor: share Go tool/args helpers, table-driven protocol routing (no behavior change)
* refactor: share GoCallAccum across Go reply parsers (no behavior change)
* refactor: share part/call extractors across Go history converters (no behavior change)
* refactor: share post_to envelope between post and go_post (no behavior change)
* refactor: pass system note into go_responses_input, drop history clone (no behavior change)
* refactor: lazy Gemini name map, shared go_tools, named tool-call cap (no behavior change)
* refactor: share shaped-tool wrapper across Go tool converters (no behavior change)
* fix: set request_timeout_s in Go test setups for rebased main (no behavior change)
* fix: expect ToolsAvailable first in tool-loop tests (matches new event order)
* simplify: reuse tool_arguments, drop restater comments (no behavior change)
---------
Co-authored-by: Julius <julius@localhost>
Co-authored-by: Jom_g <jom_g@users.noreply.github.com>
Co-authored-by: HoloPort <holoport@localhost>
* claim: #8065
* Context menu: only the row under the pointer stays highlighted while a flyout closes (#8065)
A row that owns a flyout kept its highlight for the whole of the flyout's
closing delay, so the row the pointer had settled on lit up alongside it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
* claim: #7855
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019FAWPuFoh9qxn2uS2j53QZ
* Motion interference: a check cut short by its work budget is not clear (#7855)
motion.interference stops when its work budget runs out (truncated) but
answered clear: collisions.is_empty(), so a check that reached only the first
steps read as a pass: a gear pair checked for 14 of 61 samples was "clear".
clear is now true only when every sample was checked and none collide.
The budget is an argument of interference_within, so a test can cut a check
short: B slides up to touching A, a full check is clear, and the least work
that checks a first step stops after 1 of 16 samples with clear: false (it
was clear: true before). The guide and the mechanisms skill say the same.
The issue's second request, a budget_s on motion.interference, is not here.
Refs #7855.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019FAWPuFoh9qxn2uS2j53QZ
* Motion interference: budget_s gives a check the time it needs (#7855)
The work budget is fixed, so a gear pair drawn in detail was cut short
after a few samples, or refused before its first one. motion.interference
now takes budget_s (seconds, above 0, at most 86400, as inspect.interference
allows): the check then runs until that time instead of the work budget,
looked at before each sample, and echoes budget_s in its answer. Without
budget_s nothing changes. The refusal before a first step now names the
budget that ran out and suggests budget_s.
Tests: given a time, the smallest work budget (which checks no step) no
longer stops the check, through motion.interference too; budget_s 0, -1,
1e9 and "soon" are refused.
Fixes#7855.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019FAWPuFoh9qxn2uS2j53QZ
---------
Co-authored-by: Claude <noreply@anthropic.com>
Collapse All from a component or placement row folds everything under that
row and leaves the row itself open, matching the whole-tree Collapse All,
which already keeps the design row open.
* Standard Parts and Contact Set: label the rows in words, not with raw keys
The Insert Part dialog listed a size's dimensions under the library's
keys (d, head height, socket depth), and the Contact Set dialog's
checkbox read "enabled". The dimensions now take messages (Nominal
diameter, Head height, Socket depth, ...) and the checkbox reads
Enabled, translated like the rest of the dialogs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Contact Set: translate the Enabled checkbox's label
ui_kit::checkbox paints the text it is given; the callers translate it,
as Configuration Rules' Enabled does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The Interface page's strip of tab accent colours looked up its own
word-cased names ("Sheet Metal", "Surface"), which the catalogs
mostly don't carry or carry with another meaning: in German it read
"Sheet Metal" and "Flächenkörper" where the ribbon says Blech and
Fläche. It now names each tab the way the ribbon does, through the
tab's heading message, and the Form tab gets one.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Preferences: the snap-suppress choice and Direct interaction in the user's language
The Hold to stop snapping choice joined its keys with an English "or"
("Cmd or Option", the default) in every language, and Navigation's
Direct interaction setting and its note had no message. The choice now
reads through a message with the keys as placeholders, and the setting
gets its two messages.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move the new test away from where another open PR adds one
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Dialogs: 54 labels that read English in every language get messages
The construction dialogs' selection fields (First edge, Second point,
Through point, ...), the joint types (Rigid, Revolute, Slider, ...), the
Sheet Metal and Plastic rules' rows (K Factor, Flange Gap, Min Draft,
...) and a few options and prompts are passed to the label translator as
English source text that no catalog carried, so they showed English in
every language. They are added in one pending file, with the eleven
usual translations, and a test makes sure every dialog's selection
fields keep a message.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Dialog labels: German Flange Gap is Flanschspalt, as the catalog's Flansch
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The Italian catalog lacked 1350 of the English messages. This batch
translates the 719 outside the Render workspace: Preferences, the
Assistant, the help texts, command labels, materials, navigation and
the rest. It uses the catalog's own terms (Schizzo, Lavorazione, Corpo,
Giunto, Cronologia, Struttura, Risalto, Appoggio, Aggancio a scatto).
The Render workspace's 600 messages are left for a second batch while
that workspace is still being built.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Insert Part: name the part families in the user's language
The Type list of Insert Part and the Change Part Size dialog showed the
library's English family names (Socket head cap screw, Hex nut, ...) in
every language. Each family now has a message under its library id,
and a family the catalogs don't know keeps the library's name.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Insert Part: the Name field's automatic hint in the user's language
The hint was the English word "automatic"; it now takes the catalogs'
parts.automatic_name, which every language already translates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move the family-name helper and its test away from where #8005 adds its own
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Turkish: translate the 113 messages the catalog was missing
The Turkish catalog lacked 113 of the English messages: the newer
Preferences settings, the hole presets, the slicer messages, the
Assistant's start-up options and tool summaries, and the Render
workspace's stage, shots and inspector labels. They are translated in
one pending file, with the catalog's own terms.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Translate the 87 messages main added since (the navigation bar's)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Dialogs: translate the texts a few dialogs draw themselves
The joints' Play/Stop buttons, the Motion Study's Loop and the BOM's
sub-parts checkboxes, the Stitch tolerance hint, the Flange type and the
Emboss wrap note were drawn as English literals; the joint lists named
each joint's type by its Rust name (PinSlot, PointOnPlane). They now take
messages, and the joint lists use the Type box's label, translated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move the new test away from where another open PR adds one
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The Surface, Sheet Metal, Plastic, Frame, Move, Align, Assembly, Cam,
Coupling, Sweep, Part Size and sketch editing dialogs refuse their
inputs with English sentences ("select bodies first", "pick faces of
one body"). The dialog translates a refusal through the catalogs as it
shows it, but no catalog carried these 96, so they showed English in
every language. They are added in one pending file with the eleven
usual translations.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Everything outside the Render workspace, which is still being built:
Preferences and the new navigation bar, the Assistant, the help texts,
command labels, materials, hole presets, the slicer messages and the
remaining dialog labels, with the catalog's own terms.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Everything outside the Render workspace, which is still being built:
Preferences and the new navigation bar, the Assistant, the help texts,
command labels, materials, hole presets, the slicer messages and the
remaining dialog labels, with the catalog's own terms.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Everything outside the Render workspace, which is still being built:
Preferences and the new navigation bar, the Assistant, the help texts,
command labels, materials, hole presets, the slicer messages and the
remaining dialog labels, with the catalog's own terms.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch: the status in the view and the browser's editing mark in the user's language
While a sketch is edited, the view's corner said "Sketch1 — 2 degrees of
freedom" (or "fully constrained", "constraints conflict") and the
browser marked the sketch "(editing)" or "(attachment lost)", in
English whatever the language; one degree of freedom read "1 degrees of
freedom". They now take messages: the catalogs' existing degrees-of-
freedom ones and four new ones.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch status and editing mark: the sketch's default name as shown elsewhere
The browser shows a default name in the user's language ("Skizze 1")
except on the edited sketch's row, whose mark made it a different text;
the view's status kept the stored name too. Both now use the name as the
browser shows it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Motion Study, Motion Trace, the motion plot, Explode and the contact
check drew their notes and refusals as they got them ("add a key for a
joint first", "select a point on a component", "none"), so they read
English in every language. They now go through the label translator, and
the four that no catalog carried get messages.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Double-clicking a sketch row inside a component (an inserted fastener's sketch)
activates that component so the sketch is edited in its placement. Finish and
Cancel only closed the sketch, so the component stayed active and every other
body stayed faded.
Edit Sketch now remembers the component and placement active before it
activated the sketch's own (also when sketch.edit activates the component of a
face-attached sketch), and leaving the sketch makes that one active again, as
Finish already puts the history marker back. A component activated while the
sketch is open stays active.
Co-authored-by: Phanoob <phanoob@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Maya moves the camera with Alt and any button (left orbits, middle pans,
right zooms), middle drag also pans. HFSS pans with Alt+left drag and zooms
with Shift+Alt+left; its Ctrl+left orbit is the drag from empty space every
preset already has, so a Ctrl+click on geometry still adds to the selection.
The help texts are in the catalog for every locale that has the other
presets' help.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #5592
* Canvas and Decal: send the picked face, so a large design takes the image (#5592)
A click with Canvas or Decal sent only the point, so the engine searched every face of every
body for it and gave up on a large design ("face capture work limit exceeded"). Send the
picked body, face index and path as Create Sketch does; canvas.decal takes that face pick too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Canvas and Decal test: drop the check that a point alone gives up
On upstream main a sketch on a face by its point alone no longer gives up on this design, so
the test keeps only what the picked face has to do: both images land on it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #6039
* Sketch: offer an arc length dimension for a selected arc (#6039)
The arc length dimension already existed in the engine (sketch.dimension with
type arc_length), but nothing in the UI led to it: the Dimension tool and the
context menu always gave an arc its radius.
With exactly one arc selected, the sketch context menu now offers Arc Length
Dimension next to Sketch Dimension, which adds the dimension through the
existing command.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FN5eEMrqWWXqi65FcK8GXq
* test: move the arc length menu test up to avoid a merge conflict
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: sascha2101 <221609842+sascha2101@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #5517
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Right-click opens the menu despite tiny pointer movement (#5517)
A Magic Mouse moves the pointer a fraction of a point as it is clicked.
The right button orbited on any movement, and the menu needed exactly
zero movement in the release frame, so most right-clicks only nudged
the view. The right button now navigates once the press can no longer
be a click (egui's click distance and time), catching up with the
travel since the press, and a right-click is egui's click.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #7947
* Browser: reveal a placed body's row when it is picked or found (#7947)
The off-screen skip in `entry_row` keeps the reveal target's row alive so it
still has a rect to scroll to, but it matched `app.tree.reveal` (a world body
name) against `Entry::key` (the placement-local name). For a placed body those
never match, so the row returned early and the `scroll_to_rect` below it, which
does use `row_key`, was never reached. `app.tree.reveal` is then cleared at the
end of the frame, so there was no second chance.
Match the skip on `row_key` minus its `b:` prefix, as the `scroll_to_rect`
already does.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
The Box, Cylinder, Cone, Sphere, Torus and Coil dialogs labelled their
size rows with the command's parameter names (length, top_diameter,
section_size), in every language. Each row now takes a message: Length,
Top diameter, Major radius, Section size, translated as the rest of the
dialog is.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>