* claim: #6843
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PsiqyXWBFGTCX9J51bhysc
* Sheet metal: material joined in the plane of a panel is kept (#6843)
An extrude join through sheet metal went through the generic boolean:
the body grew, but the sheet model didn't, so the flat pattern, DXF and
bend table were refused, and the next sheet feature (a refold) dropped
the material ("what ... changed is not kept").
An untapered extrude join one sheet thick in the plane of a panel (the
base or a flange's panel, unfolded or folded) now joins the sheet model:
each region of its sketch becomes part of the one panel it is one piece
with, apart from every bend zone and the other panels, so the flat
pattern, its DXF and a refold keep it, and it follows its sketch's
parameters as a cut made there does. Cut-outs it covers are filled. A
join of another depth, out of every panel's plane, touching a bend zone
or another panel, or over a pocket, label, bore or perforation keeps the
generic path and its warning.
The extrude records its sketch plane and depth for a join too (its
profile grows into the material it joins, so it never counted as
untrimmed). The test that refused a joined tab beside a plain base now
refuses a block standing on the sheet instead; the tab is kept.
Fixes#6843.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A tapped hole looked for the cylindrical wall its marking lies on in
every face of every body of the design, measuring each face over all of
its body's triangles. A body is now passed over when the drill line,
widened by the hole's radius and a margin, misses its box (from the
mesh it keeps, which the wall's triangles are part of): no face there
can be the wall, so the marking stays the same.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #8151
* Opening: a design saved by 0.24 keeps the body a Join To Object runs to (#8151)
Since 0.24.1 a Join extrude To Object also joins the body it runs to
(#6424). Designs saved by 0.24.0 and earlier were built without that,
so on 0.25 the reported box swallowed its lid: the lid's sketch, the
holes, the cut and the moves after it failed.
An extent may now save `to_body_apart: true`, which leaves that body
apart. One pending format change sets it on every Join extrude To
Object of a design of format 39 or older as it opens; designs of
format 40 and later keep the join. Editing the extrude keeps the field.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Markus Kaltenbrunner <kaltenbrunnermarkus@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A pattern or mirror replays its source hole's definition, so a hole with
reference edges finds its centre again from them. That happens after the
hole is drilled: through the part, the centre is then in the opening, and
the check that the centre lies on the face refused every copy. The replay
now skips that check; placing the hole itself still makes it.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Before a cut ran, every sheet metal body of its component went through a
boolean with the tool, to make sure the cut would not take material the
flat pattern cannot keep. For a sheet the tool does not reach, that check
only found nothing: in a design of 50 sheet parts it was about an eighth
of a hole's time. Now a sheet whose box lies clear of every tool's box is
left out of the check, with the same grown boxes the cut itself uses
(#10225), so a tool that touches or grazes a sheet is still checked.
Sheets shown in another frame and Intersect are checked as before.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Convert to Sheet Metal refused a tray made with SolidCraft's own sheet metal, exported and
imported again: its walls run on past each bend's end (notch, tear and round corner reliefs), so
the edges beside the bend cylinder's edge never reach the other face's plane and the face could
not be trimmed ("a face beside a chamfer picked as a bend does not reach the face across it").
Where that edge is round or runs on along the bend's line, the face now keeps its corner and
steps down square to the other plane, a step the bend's setback takes up again; the inset drops
an edge the setback leaves shorter than the tolerance.
Test: trays_of_its_own_sheet_metal_convert_back (notch, tear and round, the walls made in one
flange feature, and notch and round with the short walls in a second one): the volume, the flat's
size and outline area and every bend's length are the source's.
Fixes#7529.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #6452
* Engine: a pattern or mirror copy of a join keeps its ends as drawn (#6452)
The copy's tool is rebuilt against the current state, which already holds
the source feature's own material; the embedded-join-end trim (#2059)
read that material as the target already filling the copy's bottom band
and trimmed it away before the copy matrix was applied. Mirroring or
patterning a join on a hollow part silently lost the floor-level slice.
Copies now skip the trim: their ends are where the source drew them.
* chore: re-trigger duplicate PR check
---------
Co-authored-by: Dahak Builds <dahakbuilds@users.noreply.github.com>
A hole placed on the side of a cone with no direction drilled along
the nearest display-mesh triangle's normal, up to 2.6° off the
surface's normal and sideways. Planes, cylinders and spheres already
gave their exact normal; cones and tori now do too, from the surface
the kernel reports for the face.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A flange picked on one segment of a stepped base outline ran the base
face's whole length: trimming the picked edge back to the bend line
merged it with the outline's own edge on that line beyond the step, and
the flange took the merged edge. It now stays off the base's edges that
already lay on the bend line, so it spans the segment picked, with no
relief there: the base is already back at the bend line. The same
outline made by cutting notches into a rectangle gave the right flange
already.
Assisted-by: LLM
* Hole: a hole moved off the material says it removes no material (#10701)
A hole whose position, parameter or sketch point moves it off the part
cut nothing and said nothing: the feature had no warning, and a
parameter sweep listed the value as fine. A hole whose tool is clear of
every body it cuts now warns "the hole removes no material" ("1 of 2
holes remove no material" for several), as a pattern does of its
copies. The check compares boxes only, from the meshes the hole already
found its face on, so a large hole pattern costs no extra booleans.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Hole: a hole wider than the part says it removes all of it (#10702)
A through hole wider than the part removed the whole body with no
message, also when a parameter widened it later. The hole now warns
"the hole removes all of Body1" when a body it cuts is gone after the
cut.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #10127
* Assemblies: a construction axis follows its component when the component rotates (#10127)
A construction axis or point built on a body now goes with that body when
Move/Copy moves or turns it. Each construction records the bodies its picks
lie on; the Move feature (not a copy) carries along every construction whose
bodies it all moved, with the same transform in the construction's component
frame. One that also reads something fixed stays where it is.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Markus Kaltenbrunner <kaltenbrunnermarkus@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
sheet.fold refused any fold whose bend zone wasn't a rectangle. It now
crosses the sheet's own outer edges, slanted or on any curve: the outline
is cut exactly across the zone, the panels keep their real outlines, and
flanges and other folds still go on them. The flat pattern's DXF keeps the
curve as drawn. A sharp corner in the zone stays refused.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #10689
Co-authored-by: Cursor <cursoragent@cursor.com>
* Edit Hole: seed the face where the hole is now (#10689)
Project the stored position onto the drilled face along the hole direction when opening the editor, matching evaluation, so OK works after a parameter moves the face.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* Holes and cuts: count the bodies a cut works on in the cost tests
A test helper, cut_work, adds a feature at the end of a design and reports
how many bodies its cut ran booleans on and which bodies it left exactly
as they were (same geometry, key and face names). A new test pins that a
cut still goes through the boolean where its tool touches a face, runs
along an edge, passes a hair beside a body or drills into a dome.
The sheet metal hole (#10226) and cosmetic thread (#10227) work reuse the
helper for their own shortcuts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Holes and cuts: leave bodies the tool cannot reach untouched without a full cut
A cut ran a whole boolean against every body of its component, and for a
body its tool does not reach that boolean only handed the body back: about
7 ms per body, so a hole in a design of 200 blocks took 1.3 s. Now a body
whose box lies clear of every tool's box is left as it is. Both boxes are
their meshes' grown past how far the exact faces can curve out from them,
the same test sheet metal cuts already use, so a tool that touches a body,
grazes it or drills into a dome still goes through the boolean. Bodies
shown in another frame, mesh bodies and Intersect (which removes the
bodies it misses) are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Cosmetic tap markings: look for their cylinder only in nearby bodies
A tap marking (a Thread feature, or the marking of a tapped hole) looked
for the cylinder at its point in every body of the design, and each
search meshes the body afresh: with many bodies and many markings that
grew quadratically (200 markings on 200 cylinders: 33 s of a 33 s run).
A body is now passed over when its box, from the mesh it keeps and
grown past that mesh's deflection, lies farther from the point than the
search itself would look (a twentieth of the body's size, plus its own
mesh's deflection). The search finds no face there, so the marking and
everything else stay the same.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A Cut To Object on a curved face ran the tool through the target and kept
the tool less the body, the piece at the start. From outside a curved wall
to its inside face that piece is the air before the wall: the cut took
nothing, or the tool lay in the curved face and the boolean failed on the
coincident faces.
Where the target's material ends at the picked face under the whole
profile (rays over the profile cross it once, material before, none
after), the tool now ends flat halfway across the gap beyond the face. It
takes what lies before the face and meets no face of the body. Otherwise
the trim is as before.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A relation given no values already started from where its parts sit
(relation_values), so the part moves no further than it must. A typed joint
(planar, revolute, cylindrical ...) added between parts another joint
already relates started its values at 0 instead, so the loop solver began
far from the current placement and could report a conflict for a joint that
agrees with how the parts are placed.
Seed every loop-closing joint, typed or relation, from the current
placement when no values are given. Joints in an open chain and joints with
explicit values keep their initial values as before.
* claim: #10852
* Circular pattern: the axis origin takes length expressions and follows them (#10852)
solid.pattern.circular with axis {origin, dir} kept its origin as plain
numbers, so a pattern about a parameter-driven point stayed where the
point was when the pattern was made: the seed moved with the parameter
and the copies did not. Expressions were refused (#248).
The pattern's origin is now a point of expressions, as a Mirror plane's
origin is: numbers or length expressions, evaluated on each rebuild, and
renamed with their parameters. Plain numbers are saved as numbers, as
before. timeline.edit sets pattern.origin as expressions, and editing the
pattern in its dialog keeps them. One pending format change.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
A surface extruded from a fit-point spline is one B-rep face per span,
joined by hidden seams; the viewport picks and highlights it as one face
(Mesh::surface_patch). sketch.intersection_curve turned a face pick into
the single nearest face, so the curve stopped at that span's ends.
project3d::mesh_of now returns the picked face's surface patch and
mesh_intersection keeps the triangles of every face in it. Body sources
and Project to Surface are unchanged.
`sheet.perforate` takes an optional `sketch` (and `profiles`): the field's
outline. Holes go only inside its closed profiles, out of their inner loops,
`margin` clear of its edges, on a grid from the outline's lower-left corner.
On a folded sheet the outline maps through each panel the sketch lies on and
is clipped to it, so it never reaches a flange in the flat; a sketch on no
face of the sheet is refused. A mirrored copy mirrors the outline, and an
edit that names no sketch (the Perforate dialog has no outline) keeps it.
Without a sketch the field covers the whole sheet as before.
The feature saves `sketch` and `profiles`, a sheet's perforation its outline
in flat coordinates: one pending format note, no migration.
Assisted-by: LLM
* claim: #8822
* Tangent Plane: line it up with a reference plane, at an angle (#8822)
`construct.plane.tangent` takes an optional `reference` (a plane or planar
face) and `angle`. On a cylinder, cone or torus the plane touches where the
face turns towards the reference, turned by the angle about the face's axis,
on the side of the pick; on a sphere it is parallel to the reference. The
plane follows the reference when it moves. The Tangent Plane dialog gets a
Reference plane input and, once one is picked, an Angle field; editing the
plane brings both back. Saved as a pending format change.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9373
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Timeline folders: editing a feature in a folder leaves no stale id, and such designs open again (#9373)
timeline.redefine runs the edited feature's command on a scratch copy with the marker just after
the feature, and takes the new definition from the feature that run makes. Inside a timeline
folder that trial feature joined the folder too; it was removed from the feature list but not
from the folder, and its id, handed back, was the next feature's. Each edit left that id in the
folder (the same one, edit after edit), naming no feature or, once a later feature took it, a
feature twice, and the reader then refused the design: "damaged design: invalid feature in a
timeline folder". The trial feature now leaves the folders as it leaves the feature list.
The reader drops a folder entry naming a feature or joint the design doesn't have, or one a
folder already holds, and opens the design: folders only group the timeline, so nothing else is
lost. A folder damaged in itself is still refused.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
A mirror of placed bodies follows its source, so each frame of a drag evaluated
it and made new copies with empty mesh cells. A reflection can't share its
source's triangles, so every copy was tessellated from nothing, and so was
anything an evaluated feature after the mirror made.
Keep the first copy of a body with its source: each later copy is a proper rigid
move of it, so a world state draws them from one set of triangles. The model
keeps each copy's own lazy mesh. A feature evaluated again that gives the same
solids (an import) keeps its meshes.
A mirror of placed bodies no longer searches its copies' revolved faces when no
later feature can build on bodies. That depends on the design alone and is part
of the mirror's placement digest.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Face names: one name on two bodies is no split face (#10699)
A hole placed on the top of one of two bodies an extrude made, or on a
piece of a split body, warned "face `F2:top` was split; using the piece
nearest the picked point": the face name is on both bodies, and any
name found more than once counted as split. A name on several bodies,
one face on each, now finds the face the point lies on without the
note; pieces on one body, or a point on none of them, still say split.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Hole: look for a face name shared by two bodies on the body the point is on (#10699)
The lookup over every body counts a name on several bodies as split on
purpose (a copy's face is not a piece of the source's, see
`the_same_face_on_a_copy_is_not_a_piece_of_the_named_bodys_face`), so
the first change, which made it unique there, is undone. A hole whose
face name is found split while its point lies on one of those faces now
looks again on that body alone: one face of the name there is the face,
whole; pieces on that body still say split.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
`face_of` called `Body::faces`, which tessellates the whole body, on every
solve. Face origins went through the cached `ModelBody::joint_faces`, at
the same tolerance, but edge origins never did, so one edge origin on a
large body cost a drag or a Drive Joints step far more than the solve.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A chamfer's warning quoted the full generated names of fourteen split edges,
122,701 characters (#8541). Names over about 80 characters are now quoted by
their first 48 and last 24 characters.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
- snap_tangent_arcs : chaque extrémité d'arc prend la ligne la plus proche
- une ligne ne prend qu'une extrémité d'un arc ; l'autre passe à sa ligne suivante
- une ligne plus loin sur le contour est ignorée quand l'extrémité touche déjà une autre pièce
- tests : 4 unitaires (project_order.rs) et the_silhouette_of_conic_rounds_closes
Co-authored-by: P'tite Tête <asso@ptitetete.com>
* claim: Assemblies: use lookup maps when building the world state
* Assemblies: use lookup maps when building the world state
world_state_reusing, made again after every geometry command in a design
with components, filtered all bodies for every placement, searched the
timeline for every sketch and construction of a placed part, and searched
all bodies twice per tap and placement. It now builds the feature to
component map, the bodies of each component (in body order) and each
tap's component and body offset once per call. The result and its order
are unchanged.
On a 100-part assembly placed 5 times (800 sketches, 200 taps) one world
state drops from 124 ms to 9.5 ms (release, median).
A test-only counter of searches through all features or all bodies
checks that the work no longer grows with the sketches and taps.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Markus Kaltenbrunner <kaltenbrunnermarkus@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A join, cut or intersect without explicit targets reads where every body
of its component is placed. Its placement digest searched the model for
each of those bodies by name, then the timeline for its component and the
placements for that component's frame: O(B^2 + B*F) per feature, on every
evaluation, also when every feature comes from the cache.
The bodies, their components (Document::model_body_components) and each
component's frame are now looked up once per digest. The bodies are still
hashed in name order with the same values, so every digest stays the same
bit for bit and saved caches stay valid.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #10393
* Per-component History hiding: flag, commands, Browser menu, bar toggle (#10393)
* Carry the new History-hiding strings in the complete catalogs (#10393)
---------
Co-authored-by: Dahak Builds <dahakbuilds@users.noreply.github.com>
* claim: Canvas: store reference image data as shared text so edits stop copying it
* Canvas: store reference image data as shared text so edits stop copying it
Every edit clones the document, because the undo history still holds the
old one, and the canvas image was a plain String: each of the up to 200
undo states owned its own copy of the base64 text. Canvas::data is now
Arc<str>, like the shared STEP text, meshes and point clouds; serde's rc
feature writes it as the same JSON string, so design files stay byte for
byte the same.
A 4.2 MB image with 250 edits: peak memory 1.09 GiB -> 42 MB, run time
1.16 s -> 0.12 s.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Markus Kaltenbrunner <kaltenbrunnermarkus@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: Mesh imports: compare by pointer before comparing contents
* Mesh imports: compare by pointer before comparing contents
MeshData derived PartialEq, so the dirty check and the autosave check
walked every coordinate and index of a scanned mesh after each edit,
although the document and its saved copy share the same arrays. Compare
shared arrays by pointer first and value by value otherwise, as Points
and Colors already do. About 1.9 ms per comparison for a mesh of one
million triangles goes to nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Markus Kaltenbrunner <kaltenbrunnermarkus@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Projecting a whole body fitted its outline from mesh edges: a round wall seen
edge-on has triangles leaning a hair either way, so the outline zig-zagged
between both rims and broke into two-point pieces (lines), and seams where two
faces' triangles don't share vertices counted as open, outline edges. Faces
that face one way all over (planes, cylinders and cones about the sketch
normal) now decide for all their triangles, near vertices are welded, and a
seam between two faces facing the sketch is no outline. A STEP hook went from
159 lines to 10 lines and arcs.
A single rounded or chamfered rim of a cut curved face is a spline that only
follows its circle; read as Bezier pieces it came in as twenty splines. An
edge read as spline pieces that stays on a circle within 1e-5 of its radius,
in a plane parallel to the sketch, now comes in as that arc.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Frames: test that a renamed cut-off piece keeps its name in a rebuild from the features alone (#10438)
solidcraft-cli eval rebuilds a saved design without its saved model; the
rename made after a cut then lands on the first line's member. The test
reproduces it on main; the fix follows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Frames: a renamed cut-off piece keeps its name when the design is rebuilt from its features (#10438)
Since #10132 a cut's pieces past the first belong to the cut and the piece
that keeps the body's name to the body's own feature. Ordering the pieces
by their faces then moved the names among them but left each body's
feature behind, so the cut-off piece said the frame member feature made
it. body.rename stored its new name there, in the first line's slot, and a
rebuild from the features alone (solidcraft-cli eval of the saved file)
named member A after it. A name now moves together with the feature it
belongs to.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Find a saved model's sheets by their text, not a record built per state
Saving a design with its model cache asked, for every state, for each sheet-metal body it holds,
and built that sheet's JSON record each time only to find it written already. A design of 906
states with 300 sheets spent most of its save there. The sheet's text written straight from it
is the key now; its record is built only the first time. The records and the file are the same.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Write the model cache manifest without copying the model's records
The manifest was built as a JSON tree, which copied the whole model's records (13 to 33 MB of
records in large designs) before writing them, and a cache too big for the file wrote it again for
each try at fewer states. The manifest is now written from the records where they are, its fields
in the sorted order the tree wrote them, so the file is the same byte for byte.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A sketch holding a text whose words come from a text parameter is left
unsolved after the words change and its links then move far, with the
warning "the sketch constraints conflict (k1, ..., k10)". The same two
changes the other way round solve.
Repro: a sketch with three projected points, a construction rectangle on
them and a centred text along its far side, the text's words given by a
text parameter. Change the parameter's words, then move the source
geometry about 280 mm. Without the word change it solves; with it, the
sketch warns of conflicting constraints.
Other words are other outlines: the text is built anew with new points
and curves. The solve that walks a sketch from its stored state to a far
link move (and the check for sides the jump turned over) refused the two
sketches as not the same entities, and the single jump left alone failed.
`Sketch::stored_with_rebuilt_texts` gives the stored sketch as this one
has its entities when they differ only by rebuilt texts: the texts'
entities as they are now, every other point, radius, arc turn and
constraint value as stored. `Document::solve_sketch` walks from that, so
the points pair up again. Sketches that differ in anything else are
refused as before.
Test: sketch_project::side_tests::a_label_follows_its_source_far_after_a_text_parameter_gave_it_other_words
(fails without the fix with the conflict warning above).
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Fold the pending design format changes (align-axis-mask, equation-variable-unit, link-file-digests) into format 46.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* feat(engine,ui): allow restricting point-to-point align to chosen axes (#9946)
* style: apply cargo fmt across crates
* fix(clippy): use filter then map instead of bool::then in filter_map
* claim: #9333
* Equation curves: the variable can carry its range's unit (#9333)
A curve can set variable_unit "range" (or tick the new box in its window): the
variable is then a length, so a ^ 2 / (4 * t) is a length when a is one. Start
and end must both be lengths, or one a length and the other a plain number in
the design's unit; anything else is refused. Saved only when set, as one pending
format change; without it curves behave as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>