Continues the gradient + property-panel polish from the previous
commit and rounds out two new flows the TS app already has:
Gradient stops + effects:
- ColorTarget gains GradientStop(i) + EffectColor(i); HSV picker
preserves alpha across hue/SV drags so a transparent stop stays
transparent. Hex pill stays 6-char; alpha is reattached at commit
and the swatch sits on a 2x2 alpha checker so #00000000 reads as
transparent rather than empty.
- Effects section reflowed into card-style blocks (image #9 spec):
title + minus, X/Y and Blur/Spread 2-col grids, color row with
swatch + rgba(...) text; clicking the swatch opens an HSV picker
bound to that effect index via SetEffectColor.
- Press dispatch on both hosts anchors picker overlays at the
clicked y so they pop adjacent to the swatch instead of the top.
Image + SVG import (toolbar + Fill section "图片" row):
- New FileAction::ImportImageOrSvg / PickFillImage; persistence_image
pops rfd, decodes raster as data: URL, inserts an Image node or
rewrites the selected node's primary fill.
- ImageNode actually renders on the canvas: NodePayload + SceneNode
carry image_src, canvas_viewport_paint.rs decodes the data URL
once and hands raw bytes to RenderBackend::draw_image with a
src-hash cache id. Grey placeholder paints only when decode fails
so transparent PNGs don't get a grey matte underneath.
- SVG import ported to TS-parity (packages/pen-engine svg-parser):
recursive <g> tree walk with inherited fill/stroke/style="...",
viewBox-aware scaling with maxDim cap, multi-subpath split, raw
d preserved on PathNode. Imports land wrapped in a Group named
after the source file.
Locale-aware first run:
- settings_io detects the OS locale (LC_ALL/LANG/LC_MESSAGES with
zh-Hans/zh-Hant heuristics) and seeds editor_ui.locale before
settings.json is read; persisted user choice still wins.
- macOS bundle declares CFBundleLocalizations + AllowMixedLocalizations
so NSOpenPanel / NSSavePanel render in the same language as the
rest of the chrome.
Web host kept exhaustive across the new variants (PickFillImage,
OpenEffectColorPicker, ColorTarget::EffectColor, GradientStop). Two
new files: persistence_image.rs (file-pick handlers, ≤120 lines) and
svg_path_data.rs (path-d tokenizer + bbox + normaliser, split from
svg_import.rs to stay under the 800-line cap). 277 op-editor-core
tests pass.
Codex stop-time review caught a regression: `generate_design_code`
returned the LLM response verbatim, missing the TS
`extractHtmlFromResponse` post-processing (design-code-generator.ts
L54-87). Without it, the common LLM output shape — ```html
<!DOCTYPE html>... ``` — leaks fence markers into
`extract_structure_summary`'s downstream regex scan.
Port the 4-stage fallback chain:
1. ```(html)?\s*\n?…\n?``` fence — if inner contains
<!DOCTYPE or <html, return inner trimmed.
2. Trimmed response itself starts with <!DOCTYPE / <html → return as-is.
3. Case-insensitive find <!DOCTYPE…</html> → return slice (original case).
4. Wrap bare content in default <!DOCTYPE html><html lang=en>… scaffold.
8 new tests cover each stage + edge cases (empty, bare text,
case-insensitive doctype, fence without HTML marker, leading whitespace).
Existing `generate_design_code` tests still pass — the scripted
`<html><body>...</body></html>` outputs go through stage 2 verbatim,
identical to the old behaviour.
Three independent CI-unblockers grouped to keep history clean:
1. op-host-desktop/chat_orchestrator.rs — add `visual_ref_enabled: false`
to the DesignRequest literal so the workspace builds after S4 added
the field. Same stub-plumbing pattern as the S3b-2 `concurrency`,
S3b-4 `append_context`, S3c `validation_enabled` additions. Real
intent-gate routing is task #27.
2. op-editor-core/render_backend.rs::fill_round_rect_radial_gradient —
#[allow(clippy::too_many_arguments)] on the trait default. The
radial-gradient hook carries (rect, radius, stops, cx_frac, cy_frac,
radius_frac, opacity) per the TS pen-renderer contract; that's the
data shape, not refactorable without splitting the trait.
3. cargo fmt --all catch-up on 6 other files (op-editor-core /
op-editor-ui / op-host-native) that pre-dated this branch and were
never run through rustfmt. Pure whitespace, no semantic change. Lets
`cargo fmt --all -- --check` pass workspace-wide.
`cargo clippy --workspace --all-targets -- -D warnings` + `cargo build
--workspace` + `cargo fmt --all -- --check` all clean. Tests across
op-orchestrator (574+1) / op-design-lint (149) / op-mcp (144) /
op-editor-core (273) / op-host-desktop (96) all green.
Three clippy warnings surfaced by the final-review workspace clippy
pass that the per-task gates didn't catch:
- design_system.rs::extract_code_fence — replace manual slice
`&after_open[4..]` with `strip_prefix("json")`. Same behaviour,
cleaner intent, satisfies clippy::manual_strip.
- design_system_tests.rs + visual_ref_tests.rs — flatten the inner
`#[cfg(test)] mod tests { ... }` wrapper. The files are already
wired via `#[path = "..."] mod tests;` in their parent modules
(design_system.rs / visual_ref.rs), so the inner wrapper was double
nesting (clippy::module_inception). Matches the established
pattern in concurrent_tests.rs / plan_repair_tests.rs / run_tests.rs.
574 tests still pass. clippy + fmt + cargo build --workspace clean.
Port executeVisualRefOrchestration from visual-ref-orchestrator.ts:45-166.
5-stage flow: design-system seed → HTML codegen → screenshot → enhanced
prompt → Orchestrator::run. Emits VisualRefStarted/DesignSystem/HtmlGenerated/
ScreenshotReady/Fallback progress variants. Falls back to plain Orchestrator::run
on empty HTML or None screenshot. Abort checked before each stage.
S4 B2: port three HTML helper functions from the deleted TS visual-ref
pipeline. generate_design_code uses design-code + design-principles skills
as system prompt. extract_structure_summary uses a hand-rolled byte scanner
(no regex crate dep) to extract sections/headings/CTAs. build_enhanced_prompt
produces the byte-exact TS prompt template. 20 new tests; 569 total green.
Task B1 of S4. Adds three functions to design_system.rs:
- generate_design_system(): async LLM call with design-system skill as system
prompt; parses response via parse_design_system; falls back to DEFAULT on
any failure (port of design-system-generator.ts:20-29).
- design_system_to_seed_commands(): emits SetVariableColor/SetVariableScalar
for 8 palette tokens (color-*), 6 spacing steps (spacing-xs..2xl), 3 radius
steps (radius-sm/md/lg), 2 font strings (font-heading/font-body), and 6 type
scale steps (font-size-1..6) — 25 commands for DEFAULT (port of TS:134-156).
- design_system_to_prompt_context(): byte-exact port of the TS template prose
"DESIGN SYSTEM (use these values consistently):\nColors: bg ... Style: ..."
(port of TS:161-170).
11 new TDD tests; full suite 550 green (baseline was 539).
oxfmt across the workspace was failing the pre-commit gate on 8 files
that predate this branch — S1 op-design-lint fixtures + S3b-1a planner
golden + planner dump-golden TS tool — none of which had ever been run
through oxfmt. Pure whitespace/trailing-newline normalization, no
semantic changes. Lets format:check pass cleanly across the workspace.
Wires the linear/radial gradient fill type from the property panel
through to a real skia shader on the canvas — previously switching
fill type only changed the model and the panel painted hardcoded
placeholder stops. Stop hex (with alpha preserved), angle, offset %,
and per-stop HSV picker are all live.
- SceneNode + NodePayload carry a resolved gradient body; native
skia backend builds `linear_gradient` / `radial_gradient` shaders
with TS-renderer geometry (angle - 90° ellipse projection, radius
fraction of max(w, h)).
- PropertyFocus + EditContext gain GradientAngle / GradientStopHex /
GradientStopOffset; commit, snapshot, hit-test, paint, keyboard
validation, and the property-panel layout walker thread the new
variants end-to-end.
- ColorPickerState carries a per-session alpha so the gradient-stop
picker keeps transparency across HSV drags; hex pill stays 6-char
and re-attaches the stop's existing alpha at commit. Swatch paints
over a 2×2 checker so a fully transparent stop reads as such.
- skia.rs split into a spine + sibling gradient.rs to stay under the
workspace 800-line cap.
# Conflicts:
# crates/op-editor-core/src/lib.rs
Move the three S3c validation stub providers (SkippedPreValidator,
SkippedScreenshotProvider, SkippedVisionLlmClient) from cfg-test-only
test_support to a new pub module 'stub_providers'. Test code keeps
working via a pub(crate) re-export; host code can now construct a
no-op ValidationProviders bundle without re-implementing the trait
impls per host.
chat_orchestrator.rs (predecessor of task #27 'wire intent gate into
chat runtime') compiles again after S3b-2/S3b-4/S3c added required
fields + a required Orchestrator::run arg:
- DesignRequest literal grew append_context: None, concurrency: 1,
validation_enabled: false (conservative — validation gated off
until host wires real screenshot / vision LLM).
- Orchestrator::run takes a 6th &ValidationProviders<'_> arg; we pass
a bundle of the three Skipped*Provider stubs.
The whole module is marked #![allow(dead_code)] because no caller has
yet been added to chat_runtime.rs — that's #27 proper. Delete that
allow in the same PR as #27 真接线.
debug_validation_report was registered in rebuild_registry and listed
in tools/list unconditionally — only call() consulted
OPENPENCIL_DEBUG_TOOLS. A production client (flag unset) still saw the
debug tool in its catalog and could invoke it, getting a bare
ToolFailed. That leaks the debug surface into the production catalog.
Gate it everywhere instead of only at call time: rebuild_registry
registers the tool only when debug_tools_enabled(), and
tools_list_response appends the debug schema — now a separate
DEBUG_TOOL_SCHEMAS const, removed from TOOL_SCHEMAS — only when the
flag is set. A client without the flag never sees the tool at all,
matching the TS design where debug tools ship only in a debug build.
debug_tools_enabled() is promoted to pub and re-exported from op-mcp.
The tools/list catalog test now exercises both gate states (82 tools
flag-off, debug tool present flag-on).
Port the pen-ai-skills diagnostics layer to a new pure Rust crate
`op-design-lint`: 14 design-lint detectors, the detect_all aggregator,
apply_fixes / detect_and_fix, and golden parity tests against the TS
oracle. Wire it into op-mcp as the read-only debug_validation_report
tool, gated by OPENPENCIL_DEBUG_TOOLS=1.
Detectors: empty_paths, unexpected_rotation, excessive_frame_effects,
invisible_containers, text_explicit_heights, text_effect,
text_corner_radius, text_stroke, text_bg_contrast, edge_section_padding,
stacked_horizontal_padding, sibling_inconsistencies (+ check_consistency),
detect_all.
Also includes: node_util shared helpers + pen-core color/visibility
ports, node_mut field accessors, set_property issue->node mutation
dispatch, golden fixture corpus + TS dump script, structural-parity
test, a CI golden-drift guard, and the gitignore fix so the fixture
docs/ dir is tracked.
This branch's per-commit history was squashed: the original 28 commits
carried fabricated timestamps and could not be honestly reconstructed,
so the work is recorded as a single commit at its real completion time.
Property-panel input editing now supports:
- Arrow keys: Up/Down step a numeric field; Left/Right move the text caret. Caret position is a real index into the draft, so typing inserts at the caret and Backspace deletes the char before it (not just append/pop).
- Layer opacity: full-width box with the localized 不透明度 label inside on the left, value next to it, % at the right edge; clipped so a long-locale label can't bleed past the half-width box.
- Fill opacity: new SolidFillBody.opacity getter / setter on the model + PropertyFocus::FillOpacity + the 100 % box in the Fill head row is now editable end-to-end.
- Effect-param values: each Drop Shadow X / Y / Blur / Spread cell is a click-to-type input box (new effect_param_focus state, FocusEffectParam action, commit_effect_param_focus_if_any path). The − / + steppers still work alongside. Web's apply_property_action makes the focus a no-op (no keyboard path on web yet) to avoid stranding focus.
Paint polish:
- Standardised input-text baselines to + 19.0 across prefix / suffix / icon helpers, the fill / stroke / opacity / hex paints, and the export section so every INPUT_HEIGHT row reads on the same baseline.
- Icon-prefixed inputs now have 10 px left padding (matching X / Y / W / H) and the icon is vertically centred ((30 - 14) / 2 = 8) instead of sitting at y + 5.
- Fill / stroke swatches vertically centred in their hex rows ((30 - 16) / 2 = 7).
- '-' / '#' caret-aware validation in apply_text so typing them at caret 0 of a non-empty draft is now a valid edit. Native input_tests seed property_caret_pos to mirror real focus state.
Codex review iterations: poison-guard the effect-param focus on web, content-clip the layer-opacity row, fix property-panel scroll clamping in paint, and a few related safety guards across hosts.
Theme-toggle button now paints a Sun glyph in dark mode (click → light) and a Moon glyph in light mode (click → dark); the Sun icon was hardcoded before. Adds Icon::Moon (lucide crescent) and threads theme_mode into TopBar.
Bumps the casement submodule with a fix for the native macOS traffic-light reposition: idempotent absolute placement against resize, baseline invalidation on fullscreen exit, and a poison guard so a transitional re-capture can't drop the lights below their default position.
Add ValidationProviders bundle (approach c) carrying pre_validator /
screenshot / vision trait refs + system_prompt, passed as a new
parameter to Orchestrator::run() and threaded into all 3 execution
paths (sequential, dashboard, concurrent).
Hook site pattern (after run_cleanup_passes + CleanupDone, before
returning RunSummary):
if request.validation_enabled && !abort.is_set() {
run_post_generation_validation(sink, providers.*, &request, …);
}
Update every existing test caller to pass stub ValidationProviders
(SkippedPreValidator / SkippedScreenshotProvider / SkippedVisionLlmClient).
Port of orchestrator.ts:1247-1292.
New tests (run_tests_d1.rs): 8 tests covering all 3 paths × enabled /
disabled / abort-before-hook scenarios.
Port `runPostGenerationValidation` (TS L280-524) into `validation.rs`:
pre-check + node-count gate + 3-round vision loop with fix-history dedup,
abort short-circuit, and all 5 Progress emission points. Adds
`ValidationSummary { total_applied, rounds_run }`. 11 TDD tests in
`validation_tests_c2.rs` cover every gate + edge case (TC-1 through TC-11).
Port buildNodeFromSpec (TS:307-372) + the addChild branch of applyValidationFixes
into Rust. When a StructuralFix::AddChild arrives, build a PenNode from the
vision-LLM spec (frame/text/rectangle/ellipse/path), insert it via
EditorCommand::InsertSubtree under the parent, then call
auto_fix_parent_layout_after_add_child. Status-bar parents are protected.
Also fixes the B2 nit: extract_name_base now lowercases the name before
extracting the last word, matching the TS toLowerCase() call so "Nav ITEM"
and "Tab item" share the same base "item".
Icon resolution is a stub (d/iconId left None); a TODO marks the gap for
a future IconResolver trait. 11 new tests added; full suite 480 green.
Port SAFE_FIX_PROPERTIES (17 entries), is_valid_fix_value, and
is_valid_structural_fix from design-validation-fixes.ts:67-157.
Adds 65 tests covering all property kinds + structural fix shapes.
Port of orchestrator-sub-agent.ts:739-748. When a subtask carries
existing_section_labels (Some(non-empty)), appends the verbatim 9-line
APPEND MODE: block to the sub-agent user prompt, with each label
double-quoted and joined by ", " matching the TS format.
S3b-3 Task C3: branch `run()` to `run_dashboard_path` when
`should_use_dashboard_columns` is true for sequential requests.
Extracts the dashboard implementation into `run_dashboard.rs` to
keep `run.rs` under the 800-line ceiling (606 lines post-split).
- Logical-to-live id resolution: walks live document tree after
`InsertSubtree` via `collect_name_id_map`; resolves Sidebar /
Main Content / "{label} Slot" by name.
- Sets `subtask.generated_root_id` to the slot live id (ports
TS `orchestrator.ts:541` `generatedRootId` fallback).
- Sidebar subtasks → `live_sidebar_id`; others → slot live id.
- 3-attempt tier-gated retry ladder identical to sequential path.
- Post-loop: `reorder_dashboard_main_children` + `run_cleanup_passes`
on both `live_sidebar_id` and `live_main_id`.
- Removes `#![allow(dead_code)]` / `#[allow(dead_code)]` from
`dashboard_columns.rs` and `scaffold_dashboard.rs` (all symbols
now live-reachable).
- 4 new tests in `run_tests_c3.rs`: happy path verifies Sidebar /
Main Content children in live document; non-dashboard regression;
zero-node NoContent; abort-mid Aborted.
329 tests pass; clippy -D warnings clean; fmt check clean.
Add `build_scaffold_dashboard` in new sibling `scaffold_dashboard.rs`
(S3b-3 Task C2). Builds the horizontal root + 260px sidebar +
fill_container main column scaffold; calls
`assign_dashboard_main_parents` to synthesize row/slot frames and
**embeds them as nested children of main** in the same `InsertSubtree`
(necessary because `cmd_insert_subtree` remaps every node id —
references between row / slot / main would break across separate
commands). Returns
`(cmds, sidebar_id, main_id, scaffold_baseline)` where the baseline
matches `descendant_count(state, root_id)` after apply (mirrors TS
`scaffoldCounts.set(rootId, countDescendants(root))` in
`orchestrator.ts:1038-1053`).
Port of `orchestrator.ts:283-320` (createDashboardColumnFrames) +
`orchestrator.ts:954-1019` (useDashboardColumns scaffold branch).
`cleanup::count_descendants` upgraded to `pub(crate)` so the scaffold
can pre-compute the same baseline value. Existing sequential +
concurrent scaffold functions unchanged.
12 new tests in sibling `scaffold_tests_c2.rs`; 325 total green.
Faithful port of TS `normalizeOrchestratorPlan` dashboard branch
(orchestrator.ts:259-272): when `is_dashboard_like_prompt`, overwrite
each subtask's region.width via `infer_dashboard_section_width` and
clamp region.height to [inferred*0.6, inferred*1.6] (replace when ≤0).
Non-dashboard path is unchanged. Six new tests cover width rewrite,
height-kept-in-range, height-replaced-when-zero, clamp-to-max,
clamp-to-min, and non-dashboard-unaffected.
Ports TS orchestrator.ts:486-547 to Rust (Task B3 of S3b-3).
- Add `generated_root_id: Option<String>` field (#[serde(skip)]) to
`Subtask` so run.rs can record the top-level node id post-generation.
- Implement `get_dashboard_placeholder_height` — estimates dashboard
main-column height from first 2-3 rows' tallest subtasks + sidebar,
clamped to [560, 680].
- Implement `reorder_dashboard_main_children` — emits sequential
`EditorCommand::MoveNode` commands to re-sort the main column's
children back to plan order after sub-agent generation.
- Both functions live in the new split module
`dashboard_columns/height_reorder.rs` (207 lines).
- 9 new tests in `dashboard_columns_tests_b3.rs`; suite grows from
298 → 307 passing.
Port TS assignDashboardMainParents (orchestrator.ts:401-484) to Rust.
Implements §4.6: synthesizes fill_container Slot frames for single-subtask
rows, and horizontal Dashboard Row frames + proportional-width Slot frames
for multi-subtask rows (min 220 px per slot, last slot always fill_container,
available_width floor 320 px). Splits §4.6 into dashboard_columns/slots.rs
to keep all files under the 800-line ceiling; TDD test suite in
dashboard_columns_tests_b2.rs (10 tests, 298 total passing).
Implements Task B1 of S3b-3: `normalize_dashboard_main_subtasks` strips
metrics-row phrases from the main-content-container's elements, relabels
it to "Top Bar", and clamps its height to [88, 120].
`group_dashboard_main_rows` is a greedy bin-packer over non-sidebar
subtasks returning `Vec<Vec<usize>>` row groups + full_width + row_gap.
Standalone rule: width >= full_width*0.82 OR is_main_content_container.
Flush thresholds: pre-flush > 1.05, post-flush >= 0.92. row_gap =
root.gap when > 0, else 24. Exact port of orchestrator.ts:322-399.
17 new tests; full suite 288 green; clippy + fmt clean.
Implements Task A2 of S3b-3: ports `inferDashboardSectionHeight`,
`inferDashboardSectionWidth`, and `extractSidebarSurfaceColor` from
`orchestrator.ts:213-237` and `orchestrator-sidebar-color.ts` into
`dashboard_columns.rs`.
- `infer_dashboard_section_height`: sidebar→760, header→96,
metric/kpi→160, chart/revenue→320, transaction/activity/feed→320,
table/analytics/customer→340, default→160.
- `infer_dashboard_section_width`: chart/revenue→main*0.62 (rounded),
transaction/activity/feed→main*0.38 (rounded), else full main_width;
main_width = max(320, root_width-260); sidebar→260.
- `extract_sidebar_surface_color`: catalog "Sidebar Surface | #hex" table
match → inline match → design.md palette sidebar→panel→surface|card
role lookup → None (caller falls back to root fill or #0F172A).
Tests split into `dashboard_columns_tests.rs` (454 lines) via `#[path]`
to keep both files under the 800-line ceiling. 25 new tests, 270 total.
Part 1 (from C1 review): add `OrchestratorError::AllFailed(String)` variant
to replace the misused `Internal` in `aggregate_concurrent_verdict`; update
the doc-comment on that function; update `cleanup_tests_c1.rs` assertions.
Part 2 (Task C2): `Orchestrator::run()` now computes `screen_groups` +
`effective_concurrency` after planning and branches:
- `effective > 1` → N-root scaffold (`build_scaffold_concurrent_mobile`) +
`run_concurrent` + `aggregate_concurrent_verdict` + cleanup; on all-fail
calls `cleanup_concurrent_roots` and returns `AllFailed`.
- `<= 1` → the existing sequential path, completely unchanged.
Inline tests extracted to sibling files (`run_tests.rs`, `run_tests_c2.rs`)
to keep `run.rs` under the 800-line cap. 226 tests pass.
Add `aggregate_concurrent_verdict` (port of orchestrator-sub-agent.ts:319-325):
total_nodes = Σ node_count; Err(Internal(first_error)) iff all zero + non-empty
collected; partial success accepted. Add `cleanup_concurrent_roots` (port of
orchestrator.ts:1101-1158): per-root delete-if-scaffold-only; variable rollback
only when nothing survived across all N roots. 10 new tests in cleanup_tests_c1.rs.
Task B2 of S3b-2: implements `run_concurrent` — the concurrent executor
that drives screen-group workers via a tokio Semaphore (RAII permits,
FIFO-fair, cap = effective_concurrency), collects per-worker buffered
EditorCommands, replays them into the real DocSink in subtask-plan-index
order, and fan-ins Progress events via mpsc channel.
Also splits concurrent.rs test block into concurrent_tests.rs (STEP 0)
to keep both files under the 800-line ceiling, wired via
`#[path = "concurrent_tests.rs"] mod tests` (same pattern as plan_repair).
Updated run_screen_group_worker signature: now takes Arc<Semaphore> +
mpsc::UnboundedSender<Progress> instead of &mut dyn FnMut(Progress).
Added CountingLlm to test_support for semaphore-cap verification.
Added tokio (sync + rt + macros) to op-orchestrator Cargo.toml.
208 tests pass; clippy -D warnings clean; fmt clean.
concurrent.rs: 432 lines; concurrent_tests.rs: 727 lines.
Task B1 of S3b-2. Adds BufferDocSink and run_screen_group_worker to
concurrent.rs, enabling per-worker buffered execution with the 2-attempt
concurrent retry ladder (reduced=false/false → true/true), ready for B2
to wire join_all + serialized replay.
Adds `build_scaffold_concurrent` (+ mobile variant) to `scaffold.rs`.
One root frame per `ScreenGroup`, laid out left-to-right with gap 100;
per-group height = mobile ? root_frame.height (||812) : max(320, Σ region heights);
optional status-bar per group; returns `(cmds, root_ids, baselines)`.
Existing single-root `build_scaffold` is unchanged.
10 new unit tests, 193 total passing; clippy + fmt clean.
Adds DesignRequest.concurrency, group_subtasks_by_screen + ScreenGroup,
effective_concurrency, and clamp_concurrency in a new concurrent.rs module.
Faithful port of orchestrator.ts:780-810 (minus S3b-4 append gate).
All 17 DesignRequest literals updated; 183 tests green.
Part 1: replace hardcoded PLANNING_TIMEOUT / SUBAGENT_TIMEOUT constants
in prompt.rs with profile-derived timeouts from timeouts.rs:
- build_orchestrator_prompt Rich/Minimal → orchestrator_timeouts(prompt_len)
scaled by timeout_multiplier (port of TS fastTimeout=false branch).
- build_orchestrator_prompt Compact → builtin_planning_timeouts(tier)
(port of TS fastTimeout=true / builtin-provider path).
- build_subagent_prompt → sub_agent_timeouts(prompt_len, tier) scaled by
timeout_multiplier (port of getSubAgentTimeouts).
All three CallRequest fields (timeout, no_text_timeout, first_text_timeout)
are now populated; no_text/first_text are Some instead of None.
Part 2: remove stale #![allow(dead_code)] from retry.rs, plan_repair.rs,
and timeouts.rs — their callers have been wired in by C2/C3/this task.
