f0d4494409
fix(orchestrator): P7 governance guards for false-positive YES
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 21s
- Block Stage 2 YES when IMPLEMENTATION_UNITS contains tests: [].
- Prevent fallback from accepting orchestrator supplement examples as valid plans.
- Honor KEEP_OPEN/DO NOT CLOSE final-close dispositions by skipping close PATCH.
- Add final-close casual self-contradiction guard for YES bodies (allows explicit
`disposition: KEEP_OPEN_*` to pass through to Patch B).
- Inject rejected approaches from failure reports into next-round context with
BANNED_APPROACHES block (tests: [] / DOM mount without jsdom / Home.tsx toast
removal / git add -A).
Refs: #83 (governance break — reopen pending user decision)
#84 (Stage 2 round 5 slip — replay required after this fix)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com >
2026-05-26 13:05:39 +09:00
4da22adb43
feat( #90 ): IMP-56 u1-u19 catch-up before final close (post-u20 push fix)
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 20s
u1: text_overrides axis in user_overrides_io
u2: structure_overrides axis in user_overrides_io
u3: vite allowlist for new endpoints
u4: text_override_resolver
u5: Step 12 text_overrides apply in phase_z2_pipeline
u6: structure_override_resolver
u7: text_path_stamper
u8: SlideCanvas text-edit capture
u9: SlideCanvas structure-edit overlay
u10: userOverridesApi service extension
u11: designAgent types extension
u12: slidePlanUtils restore
u13: user_overrides endpoint tests
u14: user_overrides restore tests
u15: pipeline fallback tests
u16: edit-mode state + gating tests
u17: slide_base print mode CSS
u18: /api/connect endpoint (vite)
u19: /api/export endpoint (vite)
Recovery scope: 29 files (12 modified + 17 new). u20 already pushed in
9439575; this commit lands u1-u19 that were authored but not committed
before #90 was externally closed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com >
2026-05-26 06:12:13 +09:00
943957562f
feat( #90 ): IMP-56 u20 BottomActions wiring to /api/connect + /api/export (replace placeholder toasts + standalone HTML download + cel mirror connect; pure builders exported for vitest)
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 20s
Stage 2 final unit for Step 22 (user edit + export). u20 wires the previously
placeholder bottom-action footer to the u18 /api/connect and u19 /api/export
middlewares living in Front/vite.config.ts:
- BottomActions.tsx
• drops the dead `serializeSlidePlan` import (TS2305 blocker since u14;
project-wide `tsc --noEmit` now exits 0)
• exports three pure builders for vitest (no jsdom / RTL devDep needed):
buildConnectRequest(run_id, slug) -> POST /api/connect {run_id, slug}
buildExportRequest(run_id) -> POST /api/export {run_id}
buildDownloadFilename(run_id) -> "<run_id>.html"
• handleExport: POST -> blob -> a[download] click chain; toast on
success / failure / network error.
• handleConnect: derives slug via deriveUserOverridesKey(uploadedFile.name)
and PUTs to u18 cel mirror; reports assets_copied count.
• both buttons disable when runMeta is null so the UI cannot fire
requests with an undefined run_id.
- Home.tsx
• mounts <BottomActions/> in the footer with
{slidePlan, runMeta, uploadedFile, isLoading, onGenerate}.
• removes 2 of 3 placeholder `toast.info('… 준비 중입니다.')` buttons
(LeftMdxPanel MDX-edit placeholder remains — out of u20 scope).
• adds handleTextEdit (u15 wire to text_overrides axis) and
handleStructureEdit (u15 wire to structure_overrides axis) to satisfy
the SlideCanvas props introduced earlier in the u-series.
- imp90_bottom_actions.test.ts (new)
• 11 vitest specs locking the builder URL + JSON shape against u18/u19
middleware contracts. Verified 11/11 pass.
Stage 4 verification (all PASS):
• u20 vitest: 11/11
• u18/u19 endpoint vitest: 31/31
• npx tsc --noEmit: exit 0 (carry-forward TS2305 resolved)
• backend pytest (u1~u9 + u17 print mode, 9 files): 185/185
Out of scope:
• LeftMdxPanel.tsx:333 MDX-edit placeholder toast (separate unit)
• #1 / #72 / #74 / #79 / #80 / #81 / #93 closed dependencies (no re-impl)
• AI-generated HTML structure (Phase Z regression guard)
• frame swap via structure_overrides (locked to slot_order + hidden_slots)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-26 02:31:38 +09:00
ec7471ed59
docs( #1 ): IMP-01 A-6 u1~u5 zone_geometries_px runtime verification log (driver chain + 4-topology runs + schema lock + no-drift guardrail + pytest baseline gate; production source untouched, impl at 1dc81e0)
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 20s
2026-05-25 15:49:23 +09:00
4e281a20d8
feat( #93 ): IMP-55 u1~u12 frontend manual section swap detection (manual_section_assignment bool axis + drag-only marker gate + dual-axis persistence + backend manual-true gate)
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 9s
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-25 08:27:09 +09:00
9062931863
feat( #74 ): IMP-45 u1~u8 slide-level CSS override (frontmatter slide_overrides.css + --override-slide-css/--slide-css-file + idempotent Step 13 injector)
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 22s
u1 KNOWN_AXES tuple gains slide_css entry in src/user_overrides_io.py
(snake_case parity with image_overrides); round-trip test extends
to 6 axes.
u2 src/mdx_normalizer.py surfaces nested slide_overrides.css from the
MDX frontmatter into the normalize_mdx_content return dict; absent
key -> {}, non-string css drops. 4 unit cases in tests/test_mdx_normalizer.py
(present / absent / non-string / title-only).
u3 src/slide_css_injector.py NEW (88 lines) mirrors the
inject_image_overrides_style contract from src/image_id_stamper.py:
marker pair <!--IMP45-SLIDE-CSS:OPEN--> / <!--IMP45-SLIDE-CSS:CLOSE-->,
idempotent re-injection, </head> > <body> > document-start three-tier
fallback, empty/None -> unchanged. 8 fixtures in
tests/test_slide_css_injector.py mirror test_image_id_stamper.py.
u4 run_phase_z2_mvp1 accepts override_slide_css: Optional[str] = None;
None -> frontmatter slide_overrides.css fallback. Step 13 calls
inject_slide_css after image override injection and before the
final.html disk write, so CLI/CI/regression renders observe the same
backend artifact.
u5 argparse adds mutually-exclusive --override-slide-css TEXT (inline
CSS, <style> wrapper optional) and --slide-css-file PATH (UTF-8 read,
fail-closed sys.exit(2) on missing path / decode error / both flags
present). Resolved string is forwarded as override_slide_css kwarg.
6 cases in tests/test_phase_z2_cli_overrides.py (inline / file / both
/ missing / non-utf8 / neither).
u6 samples/mdx_batch/04.mdx frontmatter gains slide_overrides.css
block (verbatim of the former MDX04_DEFAULT_OVERRIDE_CSS constant,
no sample/frame gate). Subprocess smoke in
tests/test_phase_z2_slide_css_smoke.py verifies the marker pair and
CSS substring land in final.html.
u7 Front/client removes the sample/frame-gated frontend-only injection:
Home.tsx drops the MDX04_DEFAULT_OVERRIDE_CSS constant and the
sample==="04"+frame==="process_product_two_way" branch (-28 lines);
SlideCanvas.tsx drops the iframe contentDocument.head injection of
that prop (-14 lines). Live preview now reads backend final.html only.
u8 tests/regression/fixtures/89a_pre_baseline_sha.json 04.mdx entry
resyncs to the live SHA ddb6bf2f... / 28042 bytes (overwrites the
earlier 5-byte-drift d02c76fd... / 28047). Other entries untouched.
Note: 01.mdx baseline drift (ad6f16a3... / 29089 -> live f26a7fac...
/ 29084) predates this branch and is split to a follow-up issue per
the closed-issue fresh validation rule.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-25 03:26:03 +09:00
b4be6c1cd0
feat( #72 ): IMP-43 u1~u8 --reuse-from incremental rerun (Step 0/1/2/5/6 reuse + Step 7+ re-execute)
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 25s
u1 argparse --reuse-from PREV_RUN_ID + post-merge fail-closed guard (rejects
layout/zone_geometry/zone_section/image override axes by name; only
--override-frame is preserved).
u2 src/phase_z2_reuse_snapshot.py — JSON-only Step 6 snapshot with mdx_sha256
integrity key and {value, source_path, upstream_step} provenance per axis
(pickle forbidden per Stage 2 guardrail).
u3 _write_reuse_snapshot at the Step 6 boundary; soft-fails to stderr without
aborting the seed run.
u4 prev_run_dir RO copy of step00/01/02/05/06 + _reuse_snapshot.json into
new run_dir, state rehydration, reuse marker, frame-override application on
restored units, Step 7+ resume.
u4b fail-closed for missing prev_run_dir / missing/corrupt/invalid snapshot /
mdx_sha256 mismatch / accidental new==prev write, with value+path+upstream
diagnostics per axis.
u5 reuse_from Optional[str] threaded through run_phase_z2_mvp1 signature and
CLI dispatch; default None preserves byte-identical pre-IMP-43 behavior.
u6 Front /api/run optional reuseFromRunId forwarding (vite.config.ts +
designAgentApi.ts + run_pipeline_reuse_from.test.ts).
u7a fast CI equivalence (1 mdx × 1 layout × 2 frames); step13 whitelist =
run_id/timestamps/prev_run_id only. u7b 3 layouts × 3 mdx × 32 frames
sweep gated by pytest.mark.sweep (registered in pyproject.toml; default CI
must use -m 'not sweep').
u8 scripts/measure_reuse_savings.py argv-driven A/B/C harness with frame
pin self-discovery + seed-time exclusion; status board §8 TBD anchor
(issue-body 50-70% / 10-20s→3-8s claim explicitly unverified, not mirrored).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 22:44:27 +09:00
8648a468d9
feat( #69 ): IMP-40 u1~u6 frame contract label_default placeholder/fallback role discriminator (BIM/DX leak fix)
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 26s
- catalog (frame_contracts.yaml): F18 bim_dx_comparison_table col_a/col_b
label_default_role=placeholder; F30 industry_current_status_three_col +
F31 industry_characteristics_three_col col_a/col_b/col_c forward-compat
placeholder; F33 engn_sw_three_types untouched (no label_default).
- mapper (_build_compare_table_2col): generic _resolve_label_default(col_key)
branches on <col>_label_default_role — placeholder -> '' (Figma placeholder
suppressed at runtime), fallback -> catalog literal (legacy default), unknown
-> ValueError with template_id + role_key + value. Absent role defaults to
fallback (backward compat for contracts without discriminator).
- tests (tests/phase_z2/test_imp40_label_default_role.py): u4 generic matrix
(placeholder / fallback / absent / unknown / 3-col axis) + u5 F18-reuse
non-BIM/DX synthetic rows asserting placeholder labels emit '' and BIM/DX
literal tokens do not leak.
- snapshot (tests/integration/__snapshots__/slot_payload.json): mdx 01 F18
string_slot_nonempty.col_a_label/col_b_label True -> False (u6 expected
drift from u3 placeholder -> empty string flip). slot_names + rows + title
preserved.
Verification:
- imp40_label_default_role: 6/6 PASSED
- phase_z2 sweep: 608/608 PASSED
- multi_mdx_regression: 50/50 PASSED
- cross-suite sweep: 662/662 PASSED
- BIM/DX literal grep on mapper + new test: 0 hits
- No mdx-specific branches (mdx 03/04/05 grep on mapper: 0 hits)
Guardrails: no MDX 03/04/05 hardcoding (catalog policy only); no spacing
shrink; no auto frame swap on reject; no AI call at Step 12; F33 untouched.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 18:53:20 +09:00
028042aaa9
feat( #68 ): IMP-39 u1~u8 ranking_sort_policy single-source + backend↔frontend label-priority mirror
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 23s
u1: templates/phase_z2/catalog/ranking_sort_policy.yaml — single-source policy
(label_priority asc {use_as_is:0, light_edit:1, restructure:2, reject:3}
+ confidence desc + v4_rank asc tie-break).
u2: src/phase_z2_pipeline.py — apply_ranking_sort helper + lookup_v4_match_with_fallback
applies policy AFTER IMP-38 raw-window selection (raw default_window + usable_count
preserved on RAW all_judgments).
u3: src/phase_z2_pipeline.py — _build_application_plan_unit forwards ranking_sort_policy
+ sorted_candidate_evidence into Step 9 payload.
u4: Front/client/src/services/designAgentApi.ts — frame_candidates builder reads
unit.sorted_candidate_evidence + unit.ranking_sort_policy first; local LABEL_PRIORITY
retained only on warn-fallback path.
u5: tests/test_ranking_sort_policy.py — pure permutation coverage (sample-agnostic).
u6: tests/phase_z2/test_label_priority_synthetic.py + fixtures/ranking_sort_policy/
synthetic_divergence.yaml — low-conf use_as_is behind high-conf restructure.
u7: tests/phase_z2/test_imp39_mdx04_env_toggle_e2e.py — samples/mdx_batch/04.mdx with
AI_FALLBACK_ENABLED=off; backend selected_v4_rank == frontend frame_candidates[0].
u8: tests/phase_z2/test_imp39_corpus_audit.py — real corpus sweep over
tests/matching/v4_full32_result.yaml (10 MDX sections); section IDs loaded
dynamically (RULE 0 / RULE 7 sample-agnostic).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 17:12:07 +09:00
2e3747c5ab
feat( #88 ): IMP-88 u1~u7 Step 17 retry chain — layout_adjust + image_fit + frame_internal_fit_candidate executors + dispatcher + entry
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 23s
Step 17 salvage dispatcher previously only ran the 3 actions in
_SALVAGE_FAIL_BY_ACTION (cross_zone_redistribute / glue_compression /
font_step_compression). Any next_proposed_action outside that set hit
salvage_terminal_action and dropped through, so visual_check aborted on
layout_adjust / image_fit / frame_internal_fit_candidate cascades.
u1 — router data surface (src/phase_z2_router.py)
- ACTION_BY_CATEGORY: image_aspect_mismatch -> image_fit (new row),
frame_capacity_mismatch -> frame_internal_fit_candidate (was
frame_reselect).
- ACTION_IMPLEMENTATION_STATUS: layout_adjust / image_fit /
frame_internal_fit_candidate flipped MISSING -> IMPLEMENTED with
inline IMP-88 rationale.
u2 — failure_router cascade surface (src/phase_z2_failure_router.py)
- FAILURE_TYPE_DESCRIPTIONS + SALVAGE_FAILURE_TYPE_BY_ACTION extended
with layout_adjust_insufficient / image_fit_insufficient /
frame_internal_fit_insufficient producers.
- NEXT_ACTION_BY_FAILURE + NEXT_ACTION_RATIONALE +
NEXT_ACTION_IMPLEMENTATION_STATUS rows added; cascade chain becomes
font_step_compression -> layout_adjust -> frame_internal_fit_candidate
-> frame_reselect -> details_popup_escalation (#64 terminal).
u3~u5 — planners + apply helpers (src/phase_z2_retry.py)
- plan_layout_adjust / apply_layout_adjust_layout_css with
_layout_swap_priority across 8-preset LAYOUT_PRESETS (preset switch,
no shared-margin shrink per Phase Z spacing direction).
- plan_image_fit / apply_image_fit_css scoped to frame slot using
existing classifier image_event payload (object-fit + max-w/h
derivation).
- plan_frame_internal_fit_candidate / apply_frame_internal_fit_candidate_css
stays inside declared frame contract envelope; emits infeasible path
when envelope is absent.
u6~u7 — pipeline wiring (src/phase_z2_pipeline.py)
- _SALVAGE_FAIL_BY_ACTION extended; _attempt_salvage_chain gains
layout_adjust distinct-render branch + frame_internal_fit_candidate
CSS-overlay branch + loop cap.
- _attempt_step17_image_fit_single_pass added for image_fit entry.
- §11.7.1 / §11.7.2 entry triggers wired; Step 17/18/19 artifact
refresh + note logging closes the salvage_terminal_action fall-through
for the 3 IMP-88 actions.
Tests
- New: test_router_actions_imp88.py (12),
test_failure_router_imp88_cascade.py (12),
test_phase_z2_retry_layout_adjust.py (10),
test_phase_z2_retry_image_fit.py (13),
test_phase_z2_retry_frame_internal_fit.py (13),
test_phase_z2_pipeline_salvage_imp88.py (8),
test_phase_z2_pipeline_step17_entry_imp88.py.
- Regression-aligned: test_phase_z2_failure_router_cascade.py,
test_phase_z2_step17_salvage_chain.py — pre-existing cascade +
salvage-chain assertions updated to the IMPLEMENTED surface.
Out of scope (separate axes / issues)
- details_popup_escalation terminal body (#64 ).
- frame_reselect MISSING flip (different axis).
- Step 14/16 detection refinement.
- Stage 0 mdx_normalizer integration (locked 2026-05-08).
- AI fallback activation.
Guardrails respected
- Phase Z spacing direction: layout_adjust switches preset; no shared
margin shrink.
- AI isolation contract: planners + dispatcher are deterministic; zero
AI calls in u1~u7.
- No hardcoding: routing + cascade live in router/failure_router data
rows, not inline conditionals.
- IMP-46 (#62 ) cache carve-out: untouched.
- 1 commit = 1 decision unit: u1~u7 grouped as a single IMP-88 unit.
Stage 4 verification: 7 IMP-88 test files + 2 modified regression files
PASS (Claude #12 + Codex #12 consensus YES). Full-suite sweep deferred to
a separate step.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 15:01:55 +09:00
e0c39f1bc1
feat( #73 ): IMP-44 u1~u5 layout override unknown-key guard + frontend zone_geometries validation
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 23s
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 12:12:24 +09:00
5deeb97cf6
feat( #71 ): IMP-42 u1~u5 silent fail chain diagnostics (assert + invalid-char detector + DIAG log)
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 24s
Stage 4 binding scope — diagnostic-only, fail-loud, sample-agnostic
(RULE 0 / AI-isolation contract). No production behavior change beyond
fail-loud raises on previously-silent failure classes.
u1 src/phase_z2_pipeline.py:2747-2772 — render_slide precondition assert
(template_id non-empty str + slot_payload dict), placed after the
`__empty__` short-circuit at 2740 to preserve empty-zone grid behavior.
u2 src/phase_z2_pipeline.py:2681-2710 — _scan_rendered_html_for_invalid_path_chars
helper covering src / href / url(...) values for backslash, &, '.
Invoked on partial render (2778) and slide_base assembly (2798).
u3 src/phase_z2_pipeline.py:2638-2676,2733,5509 — _emit_diag_zones_shape
shape-only [DIAG] JSON at Step 12 slot_payload emit and Step 13
render_slide entry. No env gate — silence is the bug.
u4 Front/client/src/pages/Home.tsx:388-392 — unconditional [DIAG raw overrides]
console.log on handleGenerate boundary, after flushUserOverrides() and
immediately before runPipeline.
u5 tests/phase_z2/test_phase_z2_diag_smoke_general.py — 32-frame general
smoke driven by load_frame_contracts() registry (not literal MDX 03/04/05),
parametrizes u1/u2/u3 across the full frame_contracts.yaml top-level.
Tests (Stage 4 verification PASS):
- u1 8 passed, u2 14 passed, u3 12 passed, u4 5 passed, u5 97 passed.
- Backend full regression tests/phase_z2/ 499 passed in 110.84s.
- Frontend full regression 182 passed in 1.10s.
Out of scope (separate axes):
- Path normalization / as_posix migration.
- Autoescape policy change.
- build_layout_css refactor (Stage 1 category-error rejection).
- Recovery / auto-fix on detected invalid path.
- MDX content / frame-selection / zone-composition change.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 08:28:54 +09:00
c59864eb9a
feat( #91 ): IMP-91 u2~u15 multi-mdx regression CI suite + status-board auto-update
...
Multi-MDX Regression (IMP-91) / multi-mdx-regression (push) Failing after 31s
- u2~u5: tests/integration/test_multi_mdx_regression.py — MDX_SET=(01..05)
cached integration runs + status/structural/visual snapshots +
full_mdx_coverage assertion (9 snapshots populated for 01-05).
- u6~u11: F0 normalize / F1 V4 ranking / F2 slot_payload /
F3 classifier-only AI / F4 layout / F5 final.html axis per MDX_SET.
- u12: pyproject.toml — pytest-json-report>=1.5 in dev extras.
- u13: .github/workflows/multi-mdx-regression.yml — pytest+artifact CI.
- u14: scripts/update_status_board.py + tests/scripts/test_update_status_board.py
— idempotent JSON marker updater (3 unit tests pass).
- u15: PHASE-Z-PIPELINE-STATUS-BOARD.md — 30 F0-F5 × mdx01-05 markers
initialized `?` + workflow wiring.
Stage 4 verify: 59/59 PASS targeted (smoke 6 + updater 3 + integration 50),
386/386 PASS regression umbrella, 0 failures.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 07:01:58 +09:00
6aa7564509
feat( #91 ): IMP-91 u1 non-VP subprocess smoke mdx01/02 parametrize
...
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 02:18:17 +09:00
b1bbe27c38
feat( #89 ): IMP-89 89-a u1~u5 Layer A render path activation (B4→mapper source-of-truth switch, default-OFF flag)
...
PHASE_Z_B4_MAPPER_SOURCE env flag (default OFF) switches slot_payload
source-of-truth from legacy mapper-only / V4 rank-1 to B4 PlacementPlan
.selected_template_id at the single switch site in the runtime loop.
OFF preserves final.html SHA byte-equivalence (u4 parity guard, mdx 01-05).
ON requires Layer A render-active path; BLOCKED exits on B4 no-cover
and on B4-selected FitError (IMP-87 honesty gate pattern — NO silent
fallback). Distinct from PHASE_Z_B4_GATEKEEPER (mismatch render-skip).
Units (1 commit = 1 axis per Stage 1 scope_lock):
u1 — _b4_mapper_source_enabled() flag reader (default OFF)
u2 — _select_mapper_template_id() selector wired at the switch site
u3 — _b4_mapper_source_blocked_exit() for b4_no_cover / b4_selected_fit_error
u4 — render SHA parity regression (tests/regression/ baseline mdx 01-05)
u5 — slot_payload byte-equivalence (matches_mapper=True axis, mdx 01-05)
Targeted 89-a suite 63 PASS; Phase Z regression 323 PASS; IMP-87 mirror
20 PASS. Demo activation via .env only (no vite.config hardcoding).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-24 00:33:28 +09:00
896f273ffa
feat( #92 ): IMP-92 u1~u5 AI fallback config validation (model ping + operational error classification)
...
Replaces #84 UI-noise removal plan with positive operational-alert contract.
Five-axis stack lands together: (1) default model literal moved to current
Opus-family ID, (2) Anthropic SDK error classifier mapping exceptions to
quota/billing/auth/other, (3) api_error_kind plumbed through ai_repair_status
summary + per-record retention, (4) Step 0 preflight ping gated under
ai_fallback_enabled (default OFF preserved) with fail-fast on invalid
model/key, (5) frontend formatter rewritten to surface only operational
quota/billing/auth toasts (non-operational paths return null per
feedback_auto_pipeline_first silent-pipeline policy).
u1 - default model literal claude-opus-4-6-20250415 -> claude-opus-4-7
(src/config.py + tests/test_phase_z2_ai_fallback_config.py lock mirror)
u2 - classify_operational_error type+status_code dispatch + Step 12
api_error_kind stamp on except path (src/phase_z2_ai_fallback/client.py
+ src/phase_z2_ai_fallback/step12.py + tests/phase_z2_ai_fallback/test_step12.py)
u3 - _summarize_ai_repair_status aggregates api_error_kinds {quota,billing,
auth,other}; error_records[i].api_error_kind retained per-record
(src/phase_z2_pipeline.py + tests/test_imp47b_failure_surface.py)
u4 - _run_step0_ai_preflight + Step0PreflightError; preflight only fires
when ai_fallback_enabled=true; one-token ping; invalid key/model =>
setup failure before Step 1 (src/phase_z2_pipeline.py +
tests/phase_z2/test_pipeline_step0_preflight.py NEW)
u5 - AiRepairStatus.api_error_kinds? interface + formatAiRepairHumanReview
Message rewritten: operational quota/billing/auth -> Korean copy
verbatim from issue body (tie-break quota -> billing -> auth);
validation/coverage_violated/unsupported_kind/generic-other/legacy
payload -> null (Front/client/src/services/designAgentApi.ts +
Front/client/tests/imp47b_human_review_toast.test.tsx)
Guardrails respected:
- feedback_demo_env_toggle_policy: default OFF preserved; preflight skipped
when ai_fallback_enabled=false (test_preflight_skipped_when_disabled
asserts anthropic.Anthropic() not called).
- feedback_auto_pipeline_first: non-operational AI failures stay silent;
only quota/billing/auth reach user toast.
- feedback_ai_isolation_contract: AI remains fallback-only; no normal-path
migration; MDX preserved.
- project_imp46_carveout_caveat: cache_key/fingerprints fields untouched on
every record; no overlap with #62 cache region.
- feedback_no_hardcoding: zero MDX-sample-specific literals; classifier
dispatch by SDK type, not by string parsing.
- feedback_artifact_status_naming: operational toast scoped to alert axis,
not overall PASS signal.
Tests:
- Targeted u1+u2+u3+u4: 63 passed
- u5 vitest (Front/): 10/10 passed
- tests/phase_z2_ai_fallback dir regression: 240 passed
- tests/phase_z2 dir regression: 323 passed
- IMP-92-adjacent (-k "imp47b or ai_fallback or preflight or step12 or step0"): 299 passed (808 deselected)
- u1 baseline lock (test_client_mock.py): 8 passed
Zero failures, zero regressions outside scope.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-23 22:07:25 +09:00
842a46144c
feat( #87 ): IMP-87 u1~u5 empty_shell honesty gate + BLOCKED exit
...
EMPTY_SHELL_NO_CONTENT overall enum + 3-marker detection (frame_template_id="__empty__"
OR label="empty_shell" OR merge_type="empty_shell") routes empty-placeholder-only
slides to BLOCKED CLI exit 1 + red final_status.html, blocking fake PASS reports
(feedback_artifact_status_naming). Coverage accounting split: legacy covered_section_ids
preserved + new content_rendered_section_ids / empty_shell_section_ids. mdx05 Case B
(zero V4 evidence) honestly classified instead of synthesizing fabricated rank-1 reject
frames. IMP-30 u6/u7 stale empty-shell PASS assertions inverted (29 tests). IMP-85 smoke
parametrize: mdx05 removed from exit-0 list + dedicated BLOCKED exit test added (4 tests).
No production behavior change for chain_exhausted Case A; no AI route activation; no
mdx-id hardcoding. 53 targeted + 76 adjacent Phase Z tests PASS.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-23 20:40:54 +09:00
c53722ad0b
feat( #86 ): IMP-86 u1~u5 placeholder zones_data + invariant guard
...
Mapper FitError handler now appends a __empty__ placeholder to zones_data
and a matching debug_zone so the surviving cardinality stays in sync with
the active layout preset's grid rows. A pre-build_layout_css invariant
guard fails fast with preset/positions/count diagnostics if drift recurs.
Per-record telemetry (adapter_needed, mapper_fit_error, provisional) is
exposed on both placeholder records; authoritative slide_status.adapter_
needed_units schema is unchanged.
Closes mdx03 reject override regression: Step 12 AI router now reachable
without heights_px ValueError; default-path behavior unaffected.
u1 — FitError placeholder zones_data + debug_zone (src/phase_z2_pipeline.py)
u2 — pre-build_layout_css invariant guard (src/phase_z2_pipeline.py)
u3 — horizontal-2 normal+placeholder helper unit (test_compute_per_zone_geometry.py)
u4 — mdx03 reject override → Step 12 integration + default regression
u5 — placeholder telemetry surface (adapter_needed/mapper_fit_error/provisional)
Tests:
- u3 helper: 7 passed (0.06s)
- u4+u5 integration: 2 passed (7.87s)
- Phase Z2 + AI fallback regression: 544 passed (66.28s)
- Broader sweep (excl. matching/pipeline heavy): 1066 passed (96.12s)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-23 18:25:14 +09:00
cacc5b30db
feat( #85 ): IMP catalog builder invariant + VP runtime gate (u1~u7)
...
- u1: BuilderMissingError(FitError) — narrow exception aligned with pipeline catch
- u2: load_frame_contracts catalog invariant + VP skip + CatalogInvariantError
- u3a: audit CLI I1~I3 (partial existence / declared builder / registry membership)
- u3b: audit CLI I4 (slot_payload refs vs declared/generated payload keys)
- u4: lookup_v4_candidates VP filter (lookup_v4_all_judgments raw telemetry untouched)
- u5: catalog invariant regression coverage + temp non-VP failure fixtures
- u6: mdx04 VP routing fixture tests (sw_dependency_four_problems excluded from live)
- u7: tests/conftest.py env isolation + mdx03/mdx04/mdx05 subprocess smoke
Targeted 74 PASS (12.31s). Full regression 1063 PASS (87.70s). Audit CLI clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-23 16:56:38 +09:00
d9d338416a
feat( #62 ): IMP-46 cache fingerprint forwarding u1~u4 (router kwarg + step12 forward + 8 scenarios)
2026-05-23 08:53:22 +09:00
f3ef4d917c
feat( #64 ): IMP-35 details_popup_escalation u1~u10 + Stage 3 R7 anchor re-pin
...
Land the production + test surface for the Step 17 cascade POPUP terminal
(DETERMINISTIC -> POPUP -> AI_REPAIR -> USER_OVERRIDE) per Stage 2 plan R2.
u11 (baseline-red invariance gate) was already landed in 7c930317c930317c930317c93031#4 R2 plan, Codex #5 R2 YES
Stage 3: Claude #86 (R7 anchor re-pin), Codex #87 YES
Stage 4: Claude #88 R1, Codex #89 R1 YES
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-23 07:36:57 +09:00
7c93031f9b
feat( #64 ): IMP-35 details_popup_escalation u11 baseline-red invariance gate
...
Add a test-only invariance gate that locks the pre-existing four-test red
baseline so IMP-35 cannot silently grow the red surface while in-flight.
u11 does NOT fix the four reds — Stage 2 follow_up_candidates tracks the
actual repair as a separate issue. u1~u10 production work remains in the
worktree and is explicitly out of this commit per Stage 3 R7 carve-out.
Frozen registry (IMP35_BASELINE_RED_NODE_IDS, set semantics):
1. tests/test_imp47b_step12_ai_wiring.py
::test_mixed_units_classified_by_route_and_provisional_flag
2. tests/test_imp47b_step12_ai_wiring.py
::test_reject_provisional_unit_reaches_router_short_circuit
3. tests/test_imp47b_step12_ai_wiring.py
::test_step12_ai_repair_artifact_writes_json_serialisable_records
4. tests/test_phase_z2_ai_fallback_config.py
::test_ai_fallback_master_flag_default_off
Gate semantics (subprocess pytest, set comparison):
- All 4 node ids resolve to collectible pytest items
(rename / delete is caught up front).
- Broader baseline-area sweep across the two registry files yields
EXACTLY 4 FAILED and 0 ERROR, with FAILED set ≡ registry.
- A new red in the baseline area flips count above 4 OR introduces a
FAILED id outside the registry; either branch fails the gate.
- Cross-lock test ensures registry node ids cannot point outside the
declared area-files inventory.
AI isolation contract (feedback_ai_isolation_contract):
Gate body uses stdlib only (subprocess + re + ast). An AST self-verify
test rejects `anthropic` imports and `route_ai_fallback` references in
this file, structurally preventing AI routing inside the gate.
Stage 4 verification (HEAD c1df656#30 + Codex #31 R7 YES; Stage 4 Claude #32 + Codex
#33 R1 YES.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-23 04:13:54 +09:00
c1df656312
feat( #65 ): IMP-36 fit/rotation generalization (u1~u8)
...
Generalize Phase Z frame partial responsive fit / rotation to four canonical
F13/F14/F20/F8 family partials. Surface = 13 canonical partials; 19
builder-only contracts remain explicitly out of scope.
u1 test_imp17_comment_anchor: re-pin L570->L578 (restructure+IMP-17),
L571->L579 (IMP-29 -> IMP-47B supersession). Stage 1 red baseline gate.
u2 frame_contracts.yaml: add rotation_eligible (P1) + body_fit_pattern2 (P2)
bool axes on 13 partial-backed contracts. P1 True: F13/F14/F20/F8 (4).
P2 True: F23 + P1_set (5). F29 columns[1].body_parser column_plain ->
column_with_transform (P3 parity).
u3 test_imp36_fit_rotation_generalization (NEW, 166 lines): static
parametrized assertions for P1 metadata + CQ presence, P1 opt-out
absence, P2 --max-body-lines + clamp + cqh, P2 opt-out absence, 19
builder-only exclusion.
u4 three_parallel_requirements (F13): introduce f13b-root container-name +
container-type:size + @container (aspect-ratio<1.5) rotation;
add inline --max-body-lines + body line-height clamp/cqh/calc.
u5 three_persona_benefits (F14): f14b-root P1 + P2 cqh/jinja body fit.
Persona colors (#285b4a/#445a2f/#743002) and circle SVG aspect 1/1
preserved.
u6 dx_sw_necessity_three_perspectives (F20): f20b-root P1 + P2 cqh/jinja
body fit under IMP-49 partial-fidelity lock.
u7 info_management_what_how_when (F8): f8b-root P1 + P2 cqh/jinja body fit.
u8 test_imp36_overflow_chain_self_fire (NEW, 299 lines): Selenium self-fire
harness for F13/F14/F20/F8 at aspect 1.78 vs 1.0. Asserts line-height
changes, font-size invariance across all 4 frames (no per-frame exempt),
grid columns rotate 3 -> 1, OVERFLOW_CASCADE_ORDER remains 4-tuple.
Stage 4 verification (HEAD 6f1c736#4 + Codex #4
YES; Stage 4 Claude #1 + Codex #3 PASS.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-23 01:18:20 +09:00
6f1c7367e0
feat( #79 ): IMP-51 image_overrides axis (u1~u11 backend stamp+CLI+CSS inject + frontend drag/resize+persistence + tests)
2026-05-22 21:54:38 +09:00
bd8bcf748b
feat( #81 ): IMP-54 frontend zone editing UI (u1~u4 edit-mode body-drag + emerald highlight + pure drag-math helper + vitest)
...
u1: 4 perimeter edge strips (~8px) + top-left grip chip at zone wrapper
provide an edit-mode pointer-event surface (zIndex 25) so wrapper-level
handleZoneMouseDown becomes reachable in edit mode. Wrapper stays
pointerEvents:none and iframe stays pointerEvents:auto to preserve
text-edit reachability (A8 guardrail). Resize handles (z-30) win in
overlap regions. Iframe pointer-events temporarily forced none during
drag to prevent mouseup leak.
u2: Edit-mode isSelected branch reuses selectedZoneId with emerald visual
(border-emerald-500 / bg-emerald-500/10) distinct from pendingLayout
blue, decorative-only (pointerEvents:none inherits via wrapper rules).
u3: Pure drag math extracted to slideCanvasDragMath.ts — DRAG_THRESHOLD_PX,
crossedDragThreshold(dx, dy) strict Math.hypot > 5, and clampZoneMove
pixel→fraction conversion with x∈[0, 1-w] / y∈[0, 1-h] clamp.
Resize math (makeResizeHandler) untouched.
u4: Vitest coverage (12 tests, 3 describe blocks) on the pure helper:
threshold strict boundary at (3,4)/(5,0)/(0,5), above-threshold,
negative-symmetric, clamp negative→0, max-edge → 1-w / 1-h, per-axis
independence, non-square 500×250 slide-body, return-shape {x,y} only.
Stage 4 verify: pnpm exec vitest run client/src/components/slideCanvasDragMath.test.ts → 12/12 PASS.
Scope: edit-mode UX only. No HTML text modification, no automatic frame swap, no MDX touched.
Depends on: #9 IMP-09 (--override-zone-geometry backend wire), #80 IMP-52 (user_overrides.json zone_geometries persistence).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-22 13:35:34 +09:00
9388e25e76
feat( #80 ): IMP-52 user_overrides.json persistence (u1~u10 backend + frontend + tests)
...
4-axis MDX-stem keyed persistence so layout / zone_geometries / zone_sections / frames
survive across `/api/run` sessions. Auto-restore on MDX reopen; CLI > file precedence
on backend pipeline entry; 300ms-debounced PUT flushed before Generate.
u1 src/user_overrides_io.py — load/save/validate_key (MDX-stem regex), 4-axis schema,
miss={}, corrupt warning+{}, atomic tmp+rename, foreign-key preserve.
u2 src/phase_z2_pipeline.py — post-argparse fallback fills only missing axes.
u3 Front/vite.config.ts — GET /api/user-overrides/:key (200 {} on miss, 400 traversal).
u4 Front/vite.config.ts — PUT /api/user-overrides/:key, 4-axis allowlist, partial merge.
u5 Front/client/src/services/userOverridesApi.ts — typed get/save + flushUserOverrides
with 300ms debounce and mutated-axis partial payloads.
u6 Front/client/src/pages/Home.tsx + slidePlanUtils.ts — restore on MDX upload (non-frame
axes immediately, frames remapped post-loadRun unit_id → region.id).
u7 Home.tsx — persist on 4 mutation handlers (section drop, layout select, zone resize,
frame select); zone_sizes and Generate excluded.
u8 tests/test_user_overrides_io.py — round-trip, unknown-key passthrough, missing/corrupt,
invalid keys (26 tests).
u9 tests/test_user_overrides_pipeline_fallback.py — per-axis fill, CLI-wins, no-file noop,
corrupt warning+skip (16 tests).
u10 Home.tsx + user_overrides_write.test.ts — await flushUserOverrides() before runPipeline
in handleGenerate try-block head; source-pattern regression assertions (20 → 22 tests).
Backend pytest 42/42 green. Frontend vitest 113/113 green (endpoint 42 / restore 21 /
service 28 / write 22). HEAD baseline ee97f4f; no spillover to phase_z2 templates /
families / frames / pipeline orchestration outside the IMP-52 surface.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-22 11:47:11 +09:00
ee97f4fc78
feat( #77 ): IMP-48 composition planner re-split on all-reject (u1~u9)
...
Add resplit_all_reject_merges() helper in phase_z2_composition.py that
detects parent_merged / parent_merged_inferred units with label=reject
and rebuilds them as per-section single units using each section's own
rank-1 V4 evidence (no frame swap, MDX raw_content preserved).
Pipeline hook fires once after Step 6 settling chain (u12/u4/empty-shell)
and section_assignment_plan resolution, before Step 6 artifact write.
Guards: beneficial-split rule (>=1 non-reject), coverage equality, layout
cap (>4 abort), max_retry=1, section_assignment_override short-circuit.
Audit: comp_debug["imp48_resplit"] additive payload (applied, split_units,
skipped_units, post_split_unit_count, post_split_layout_preset);
selection_path="resplit_from_merge" telemetry on rebuilt singles;
layout_preset re-derived via select_layout_preset(new_units).
Tests: 39/39 PASS (composition u1~u6: 14 cases; pipeline u7~u9: 25 cases).
Scoped regression 720/6 with 6 failures isolated as pre-existing on
baseline 79f9ea5noreply@anthropic.com >
2026-05-22 05:00:07 +09:00
79f9ea5c92
feat( #78 ): IMP-49 dx_sw_necessity partial Figma provenance fix (u1~u3)
...
Replace eyeballed PROMOTED green hex (#296B55, #123328 ) with verbatim
upstream values from figma_to_html_agent/blocks/1171281198/index.html:
- border + check mark: #1d4d3e (upstream :208 -webkit-text-stroke)
- header gradient: rgb(15, 50, 30) / rgb(60, 52, 34) (upstream :54, :64)
Document .f20b__* as authoring-ordinal namespace (NOT Figma frame_id
1171281198); structural link via data-frame-id attribute. No selector
rename, no catalog edit.
Add focused regression test (tests/test_imp49_partial_figma_provenance.py)
extracting <style>-block hex/rgb/rgba literals and asserting non-whitelisted
literals exist byte-identically in upstream source. Whitelist limited to
neutrals (#fff, #1a1a1a) + shared zone-title token (#000 , #883700 ,
rgba(50,44,30,0.4)).
Scope: dx_sw_necessity_three_perspectives.html only. 19 missing partials,
.fNb__ rename, full 32-contract audit deferred to follow-up axes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-22 02:49:43 +09:00
2ef02f5f18
feat( #76 ): IMP-47B u11 frontend human_review surfacing (hunk-split from IMP-41)
...
- AiRepairStatus interface mirrors backend step20 u8 schema
- formatAiRepairHumanReviewMessage(): pure helper for the three failure axes
(error / coverage_violated / unsupported_kind) — null on success/no-AI
- Home.tsx: toast.error(aiReviewMsg) after run completion
- FramePanel.tsx: reject-click window.confirm guard ("frame 유지 + AI 재구성")
- imp47b_human_review_toast.test.tsx: 6 vitest cases (null/false/3 axes/other)
Verification (frontend node_modules junction from main worktree):
- vitest imp47b_human_review_toast.test.tsx: 6/6 passed
- vitest full suite: 19/19 passed (imp41_application_mode 13 + u11 6, zero regression)
Hunk-split rationale:
- stash@{0} (imp47b-frontend-u11-pre-rebase, captured before IMP-41 merged)
contained inline IMP-41 helpers alongside u11 changes
- HEAD already has IMP-41 helper-based implementation (buildBadgeTitle /
mergeApplicationCandidates from services/applicationMode.ts, f358604noreply@anthropic.com >
2026-05-22 00:34:32 +09:00
1186ad8ae2
feat( #76 ): IMP-47B reject-as-AI-adaptation activation (u1~u13 backend + tests)
...
- u1~u9: AI fallback infrastructure (router/prompts/schema/validator) + Step 12 hook
- u10: e2e reject chain (writes final.html with AI-repaired slot, full coverage)
- u11: frontend wiring deferred to follow-up commit (split from IMP-41 hunks)
- u12: coverage_invariant guard
- u13: cache save gate (visual_check PASS + user_approved/auto_cache) — Codex #22 verified
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com >
2026-05-22 00:19:10 +09:00
f358604fb3
feat( #70 ): IMP-41 application_mode forwarding to FramePanel V4 badge tooltip (u1~u5)
...
Forward backend Step 9 `unit.application_candidates[]` (application_mode /
auto_applicable / delegated_to) onto FrameCandidate and surface the
application_mode as a Korean consequence phrase in the FramePanel V4-label
inline badge tooltip. Deterministic frontend-only refactor; no LLM call,
no V4-label color change, no outer composedTitle change.
u1: types/designAgent.ts — add optional applicationMode / autoApplicable /
delegatedTo on FrameCandidate (legacy fixtures keep undefined).
u2: services/applicationMode.ts (new) — pure helper exporting
ApplicationMode union, APPLICATION_MODE_TOOLTIP_KR (keyed by backend
mode VALUE, NOT V4 label), buildBadgeTitle, mergeApplicationCandidates.
u3: tests/imp41_application_mode.test.ts (new) — 13 Vitest cases pinning
composite output per mode, undefined/unknown→legacy fallback, merge by
template_id, skip missing/empty/non-string keys, first-wins on dupes,
empty/null/non-array input.
u4: services/designAgentApi.ts — bridge consumes mergeApplicationCandidates
and forwards three fields onto FrameCandidate while preserving
LABEL_PRIORITY sort and TOP_N_FRAMES slicing.
u5: components/FramePanel.tsx — V4-label badge `title` now calls
`buildBadgeTitle(candidate.label, candidate.applicationMode)`;
badge color className map preserved verbatim; outer composedTitle
untouched.
Scope-qualified verification (5 files, IMP-41 axis only):
- Vitest: client/tests/imp41_application_mode.test.ts — 13/13 PASS.
- Diff↔Plan parity: 5 files match Stage 2 plan, no scope creep.
- AI-isolation contract honored: tooltip values originate from backend
enum; no frontend re-derivation from V4 label.
- No spacing/font shrink; clipping resolution stays at layout/zone/frame
layer (feedback_phase_z_spacing_direction).
Pre-existing unrelated diagnostics (BottomActions.tsx,
imp47b_human_review_toast.test.tsx) remain open on their own axes and are
not gated by this commit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-22 00:17:32 +09:00
90503cadd6
feat( #67 ): IMP-38 V4 max_rank policy formalization (u1~u3, 4 round consensus)
...
- u1: separate templates/phase_z2/catalog/v4_fallback_policy.yaml + load_v4_fallback_policy() loader
(catalog pollution prevention — Codex #1 correction)
- u2: dynamic effective max_rank in lookup_v4_match_with_fallback (3-variable ceiling min,
Codex #2 correction: min(configured, len(judgments_full32))) + 3-tier usable predicate
(status + catalog + optional capacity) + trace 8 fields (requested/default/configured_extended/
judgments_count/effective_extended_ceiling/effective_max_rank/usable_count/policy_applied)
- u3: 2 production call site cleanup (max_rank=3 removed, HEAD baseline) + tracked
Front/vite.config.ts PHASE_Z_MAX_RANK env retired + 4 regression scenarios
verified: 32 passed (IMP-38 focused scope) — IMP-05 L4 dedup / L2 schema preserved,
IMP-30 allow_provisional byte-identical, caller_override backward compat (tests)
Stage cycle (#67 , 7 round Claude + 5 round Codex):
- Stage 1: Claude #1 -> Codex #1 YES + 5 corrections
- Stage 2 r1+r2: Claude #2-#4 -> Codex #2 Q2 -> Codex #3 YES (4 round consensus LOCK 23195)
- Stage 3 U1+U2+U3: Claude #5-#9 -> Codex #6 NO 4to3 correction -> Codex #7 YES -> Codex #8 YES
- Stage 4: Claude #11 -> Codex #9 (anchor attribution nuance) -> Codex #10 readiness -> Codex #11
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com >
2026-05-21 22:14:05 +09:00
dceb10129f
feat( #63 ): IMP-34 R1 donor capacity measured bound (u1+u2)
...
Bound donor capacity in plan_zone_ratio_retry by min(static_slack,
max(0, clientHeight-scrollHeight)) when both Step 14 measured fields
are present; fall back to static contract slack when absent. Prevents
the donor from being over-allocated when full-but-not-overflowing,
avoiding a wasted Selenium rerender before cascade falls to
cross_zone_redistribute.
- src/phase_z2_retry.py: planner block L122-157 only; donor filter
(L107-112), slack<=0 gate, base_plan, greedy aggregation untouched.
Adds measured_empty_px + slack_bound_source telemetry to
donor_candidates_considered (additive only).
- tests/phase_z2/test_phase_z2_retry_measured_bound.py: 5-axis
regression (static_fallback / measured<static / measured>=static /
measured==0 excludes / filter+bool guard).
Guardrails honored: V4 rank-1 frame lock preserved, no frame_swap,
no spacing/padding/gap/line-height/font shrink, no content drop,
no MDX 03/04/05 branching, no Step 14 schema mutation. Static
fallback idempotent when measured fields absent.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-21 21:37:41 +09:00
a06dd3d4b0
feat( #42 ): IMP-04b catalog extension to 32 frames (u1~u24)
...
Extends frame_contracts.yaml from 11 to 32 contracts to match V4 evidence
(tests/matching/v4_full32_result.yaml unique template_ids), closing the
IMP-04b gap surfaced in IMP-04 (#4 ) Track A milestone.
Scope (Stage 2 24-unit plan):
- u3/u4: WIP partial absorb — app_sw_package_vs_solution (F23),
pre_construction_model_info_stacked (F9). Both promoted from
_WIP_FILES.md to frame_contracts.yaml. WIP allowlist now empty.
- u5~u11: Track A 7 frames (index.html present, contract missing).
- u12~u23: Track B 12 frames (visual_pending: true; family partial
authoring deferred — contract-first per Stage 2 plan).
- u24: BT closure gate. Adds
test_imp04b_closure_gate_v4_coverage_and_wip_empty (catalog ↔ V4
set-equal + WIP==0) and test_vp_exempt_keys_are_contracted_and_disk_absent
(vp ∩ disk == ∅). Relaxes test_contracts_set_equals_disk_families_minus_wip
to (disk - wip) ∪ vp. 32 derived from V4 evidence YAML (no hardcoding).
Closure facts (locked):
contracts = 32, v4_unique = 32, missing = [], extra = [],
wip_count = 0, vp_count = 19, vp ∩ disk = [].
Guardrails honored:
- No calculate_fit migration.
- No AI/Kei API call in per-frame work.
- No 1-2 sample hardcoding (Codex #7 generalization guardrail).
- No production refactor for tests (IMP-32 owns helper extract).
- figma_to_html / V4 / Phase Z 3-layer separation preserved.
- 1 commit = 1 IMP-04b decision unit (bundled u1~u24 per Stage 2
plan; CAT+WIP atomicity for u3/u4 preserved).
Tests: tests/test_family_contract_baseline.py 4/4 PASS.
Cross-ref: IMP-04 (#4 ), IMP-29 (#38 ), IMP-30 (#39 ), IMP-31 (#40 ),
IMP-32 (#41 ), IMP-33 (#61 ), IMP-47A (#75 ).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-21 19:39:16 +09:00
15ef7c65e9
fix( #75 ): IMP-47A mdx03 frontend execution stabilization (u1~u4)
...
u1: SlideCanvas iframe sandbox += allow-scripts (allow-same-origin preserved)
→ embedded-mode script in slide_base.html now applies html.embedded
→ standalone CSS reset deactivates inside iframe; no clipping
u2: designAgentApi.loadRun merges candidate_evidence + v4_all_judgments
+ v4_candidates via Map<template_id|id|frame_id> dedup,
LABEL_PRIORITY (use_as_is<light_edit<restructure<reject) then
confidence desc, capped TOP_N_FRAMES=6
u3: Home.handleGenerate useCallback deps = [uploadedFile, slidePlan,
userSelection, pendingZones, pendingLayout] (5-tuple, stale-closure fix)
u4: tests/manual/imp47a_e2e.md — mdx03 manual e2e spec (5 axes)
Frontend-only. Backend src/ untouched. No template/catalog edits.
Determinism preserved (no LLM in frontend merge logic).
Baseline: pytest -q tests → 623 passed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-21 14:56:56 +09:00
c864fe0479
feat( #61 ): IMP-33 AI fallback scaffolding (u1~u11, flag default OFF)
...
Frame-aware AI fallback module scaffolded under src/phase_z2_ai_fallback/
with master flag ai_fallback_enabled=False; normal-path AI call count
remains 0. AI output constrained to builder_options_patch /
partial_overrides / slot_mapping_proposal; MDX / frame_id / raw HTML /
raw CSS mutations rejected at schema layer. IMP-46 cache gate (cache.py)
raises AiFallbackCacheGateError unless visual_check_passed AND
user_approved. Step 12 wires AI repair after IMP-30 provisional payload
only; Step 17 stays blocked behind IMP-34 / IMP-35 prerequisites.
AST isolation guard forbids fallback package from importing Phase Q /
Kei / pipeline runtime symbols. Docs IMP-17 / IMP-31 bound to runtime
module surface via 11-row structural test pin (test_docs_sync.py) so
drift fails CI.
Tests: 116 fallback / 161 phase_z2 regression / 526 scoped full sweep
all passing. Existing pre-IMP-33 fixture issue in scripts/test_phase_t_*
remains untouched (out of scope).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-21 12:46:49 +09:00
c412f1ea75
refactor( #41 ): IMP-32 Step 9 application_plan helper extraction (u1~u5)
...
Pure refactor — extract inline Step 9 per-unit application_plan dict
assembly into module-level private helpers for testability. Replaces
IMP-05 Case 7 inspect.getsource() literal guard with direct helper-call
shape test. Behavior preserved: key set/order, candidate_evidence +
fallback_chain compat alias identity, IMP-06 additive plan fields,
IMP-11 D-2 markers (single _contract = get_contract(c.template_id)
bind + catalog_registered + min_height_px chain).
- u1 _application_candidates_for_unit(unit) at src/phase_z2_pipeline.py
:2829-2853 — APPLICATION_MODE_BY_V4_LABEL mapping (pure extraction)
- u2 _v4_all_judgments_for_unit(v4_all_for_unit) at :2855-2882 —
IMP-11 D-2 chain preserved literally
- u3 _build_application_plan_unit(unit, zone_plan, selection_trace,
plan_record, v4_all_for_unit, layout_preset, layout_candidates_list)
at :2885-2995 — byte-identical per-unit dict (key set + order +
value identity), candidate_evidence / fallback_chain compat alias,
v4_candidates list, v4_all_judgments, application_candidates, IMP-06
additive plan fields
- u4 Step 9 inline loop body at :4620-4658 replaced with helper call;
per-index/per-id lookups (zone_region_plans[i], v4_fallback_traces
.get(...), plan_record_by_unit_id.get(id(unit)), section_alias_by_id,
lookup_v4_all_judgments(...)) stay at call-site
- u5 tests/test_phase_z2_v4_fallback.py Case 7 rewritten to
test_build_application_plan_unit_emits_candidate_evidence_and_alias
— direct helper call with SimpleNamespace duck-typed input; asserts
candidate_evidence list identity (is), fallback_chain compat-alias
identity (is), key order (candidate_evidence before fallback_chain),
and compat-alias comment scoped to inspect.getsource(_build_
application_plan_unit)
Verification: targeted 22 passed, full pytest 408 passed (0 fail/skip),
smoke 11/11 PASS (2 pre-existing baseline SKIPs unchanged).
Cross-ref: IMP-05 (#5 ) commit 23d1b25#20 + #21 / IMP-11 D-2 marker preserved.
2026-05-21 03:17:27 +09:00
182aa7c47f
docs( #40 ): IMP-31 gate audit + activation reference cross-link
...
- u1: IMP-17-CARVE-OUT.md anchor cite :564 -> :570/:572/:575/:580/:664
- u2: new IMP-31-GATE-AUDIT.md (3-cond AND gate state + 8 issue-body axes)
- u3: backlog row + status-board section 5 cross-ref to audit doc (no verdict dup)
doc-only. no src/ templates/ tests/ touched. src/phase_z2_ai_fallback/ not created.
2026-05-21 01:59:03 +09:00
1efbf672bd
feat( #39 ): IMP-30 first-render invariant + abort bypass (2 paths)
...
Restore first-render invariant: final.html + Step 20 slide_status MUST be
written for every input where Step 0~5 succeed. Two abort paths replaced
with provisional/empty-shell synthesis; MDX content preserved, AI-free.
- u1 V4Match.provisional + lookup_v4_match_with_fallback(allow_provisional)
chain_exhausted -> synthesize rank-1 provisional (opt-in, default-off)
- u2 CompositionUnit.provisional propagation (single / parent_merged /
parent_merged_inferred constructors)
- u3 select_composition_units(allow_provisional_fill=True) last-resort
fill + _candidate_state="selected_provisional"
- u4 pipeline.py path-(a) abort guard replaced with provisional retry +
terminal __empty__ shell (no sys.exit(1))
- u5 zones_data.provisional -> slide_base.html zone--provisional class +
data-provisional + needs-adaptation badge (template-only)
- u6 compute_slide_status additive provisional_first_render_count/_units
(overall enum unchanged per IMP-05 Codex #10 D4)
- u7 regression: tests/test_phase_z2_imp30_first_render.py (28 tests) +
tests/test_phase_z2_v4_fallback.py (+5 cases)
Guardrails verified: MVP1_ALLOWED_STATUSES unchanged, no calculate_fit,
no LLM in fallback path, no MDX 03/04/05 hardcoding.
Anchor sync (Rule 13): tests/orchestrator_unit/test_imp17_comment_anchor.py
re-pinned 564/565 -> 570/571 to track V4Match.provisional shift at
src/phase_z2_pipeline.py:179-184.
Cross-ref: IMP-05 (#5 ) §5 defer + Codex #2 first-render invariant.
2026-05-21 00:40:58 +09:00
b4872ba6ce
feat( #38 ): IMP-29 frontend zone-level evidence bridge (candidate_evidence reader + types + UI)
2026-05-20 21:53:47 +09:00
265d70ed91
refactor( #28 ): IMP-28 L4 _parse_json dedup (4 modules -> src/json_utils)
...
Consolidate duplicate _parse_json helpers from content_editor.py /
design_director.py / kei_client.py (fuller form) and pipeline.py (simple form)
into shared src/json_utils.parse_json (strict superset). All 18 call-sites
preserved via `parse_json as _parse_json` alias import; no behavior change.
- src/json_utils.py (new): shared helper, fenced/plain-fence/bare-brace patterns
+ list-prefix cleanup fallback.
- tests/test_json_utils.py (new): 9 unit tests pinning parser semantics.
- src/content_editor.py / design_director.py: remove local helper +
unused `import json` / `import re`.
- src/kei_client.py / pipeline.py: remove local helper; `json` / `re` retained
(used elsewhere).
Targeted tests 9 passed; full pytest 374 passed (3 pre-existing scripts/
collection errors reproduce on baseline 909bf75
2026-05-20 20:44:19 +09:00
909bf75edc
refactor( #27 ): IMP-27 K5 catalog loader + _get_block_by_id cleanup
...
Consolidate three duplicated catalog readers and two _get_block_by_id
implementations behind a single shared module (src/catalog.py) that owns
file-read + mtime cache. All caller signatures and return contracts
remain byte-identical.
Units:
- u1 NEW src/catalog.py (76 lines): load_root_catalog / load_blocks /
get_block_by_id / get_catalog_mtime as the sole file-read +
mtime-cache owner.
- u2 src/block_reference.py: _load_catalog delegates to load_blocks
(list[dict] preserved); _get_block_by_id (no-arg) delegates to
catalog.get_block_by_id. Module-level _catalog_cache removed.
- u3 src/block_selector.py: load_catalog delegates to load_root_catalog
(root dict preserved); _get_block_by_id (catalog-injected sig
preserved) delegates to catalog.get_block_by_id. Module-level
_catalog_cache / _catalog_mtime / CATALOG_PATH removed.
- u4 src/renderer.py: _load_catalog_map and
_load_catalog_map_with_variants consume catalog.load_blocks; renderer
projection caches kept local but keyed via
catalog.get_catalog_mtime(). Per-projection invalidation keys
(_CATALOG_MAP_MTIME / _CATALOG_VARIANT_MAP_MTIME) introduced. import
yaml, CATALOG_PATH, legacy _CATALOG_MTIME removed.
- tests NEW tests/test_catalog_shared_loader.py (421 lines, 23 cases):
shared loader + 3 wrappers covering single file-read, contract
preservation, signature preservation, shared cache, private state
absence, mtime invalidation propagation to renderer projections.
Verification:
- pytest tests/test_catalog_shared_loader.py -v: 23/23 PASS in 0.13s.
- pytest tests/ -q --ignore=tests/matching: 365/365 PASS in 38.10s.
- src/fit_verifier.py, src/space_allocator.py, src/pipeline.py and
templates/catalog.yaml unchanged (git diff empty).
Out of scope:
- catalog.yaml schema/path unchanged.
- Catalog direct-read call sites in fit_verifier / space_allocator /
pipeline left for a separate follow-up axis.
- Phase Z 22-step runtime, frame_selection, light_edit/restructure
flows untouched.
Refs: IMP-27 (gitea #27 ), INSIGHT-MAP §5 K5, PHASE-Q-AUDIT §2.10
2026-05-20 19:31:26 +09:00
2896bb691c
docs( #26 ): IMP-26 J3 status pending->deferred + dual-precondition trigger
...
BACKLOG line 93 + INSIGHT-MAP line 150 (verbatim mirror per anchor sync
rule). Trigger axis now requires both Phase R' archive trigger AND
§2.1/§2.2 SoT signature unification to keep guardrail = code-removal-only.
No source files touched.
2026-05-20 18:27:55 +09:00
a71355e005
docs( #25 ): PHASE-Q-AUDIT §1 lens B-1 row candidate-file 칸 정정
2026-05-20 17:52:29 +09:00
b1897c01bc
docs( #24 ): PHASE-Q-AUDIT §1 lens A-2 row candidate-file 칸 정정
...
block_reference.py, block_selector.py 를 간접 reference (catalog 로딩 /
block 검색 패턴) 로 재분류. A-2 main = frame_contracts.yaml + frame_partials
신규 구축 (Phase Q catalog schema ≠ Phase Z) 임을 명시. §2.10 K6 + §3-A
1242 binding SoT 와 정합. IMP-22 / IMP-23 A-3/A-4 lock 패턴 추종.
2026-05-20 17:19:24 +09:00
5d23b747ff
fix(orchestrator): P5b first-line agent header strict + supplement throttle
...
Bug discovered during #24 IMP-24 K6 Stage 2 (2026-05-20):
- Codex r1, r2, r3 started with '=== IMPLEMENTATION_UNITS ===' on first line
(not '[Codex #N] ...'), so detect_agent (P0-1 strict, first-line only)
returned None.
- For non-audit issues, the P5 supplement guard was audit-only gated → silent
loop until Codex r4 happened to use correct format. 4 rounds wasted.
Verified that #21 Stage 4 had the same latent silent loop pattern
('## [Codex #1 ]' first line) — orchestrator looped through ~10 Claude rounds
before random recovery. P5b fix addresses this long-standing bug.
Patch (defensive parser-contract hardening; does not assume single root cause):
1. RULES global gets explicit "FIRST non-empty line MUST be [Claude #N] /
[Codex #N]" rule that OVERRIDES any stage-specific "body MUST contain"
constraint.
2. COMPACT_PLAN_RULE wording clarified: "body" begins AFTER the first-line
agent header. The 'body MUST contain ONLY' set no longer accidentally
permits '=== IMPLEMENTATION_UNITS ===' on line 1.
3. is_codex None supplement guard:
- audit-only gate REMOVED → fires for all issues (#24 latent loop fixed)
- Throttle: max 2 supplements per stage; on 3rd violation, orchestrator
hard-stops the issue with explicit "user action required" message
and exits run_stage cleanly
- Supplement message names both Claude AND Codex (Claude's first-line
violation also breaks downstream via Codex mimicry)
- Body-head 80 chars logged on detection failure (debugging aid)
4. Regression tests (+5 cases in test_orchestrator_core.py):
- TestDetectAgent: '=== IMPLEMENTATION_UNITS ===' first line → None
- TestDetectAgent: [Codex #N] first line + units after → 'codex' OK
- TestDetectAgent: '## ', '📌 **', '**' prefix all → None
- TestRulesAndCompactPlanFirstLineContract: RULES wording has FIRST/OVERRIDES
- TestRulesAndCompactPlanFirstLineContract: COMPACT_PLAN_RULE has carve-out
Cosmetic side effect (accepted): Claude's '📌 **[Claude #N] ...**' or
'## [Codex #N] ...' decoration prefixes will fail detect_agent. Agents
will drop decorations from line 1; line 2+ can still use them.
Out of scope (NOT included to keep regression risk low):
- detect_agent function logic UNCHANGED (P0-1 strict preserved)
- consensus parser UNCHANGED
- stage loop structure UNCHANGED
- git/Gitea retrieval logic UNCHANGED
- audit-only mode P4/P4a guards UNCHANGED
- pre-post comment validation (future axis, larger refactor)
Total: 131/131 pytest pass (126 prior + 5 new).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-20 17:01:24 +09:00
447e702520
docs( #23 ): PHASE-Q-AUDIT §1 lens A-3/A-4 html_generator 칸 정정
...
§1 lens A-3 / A-4 candidate-file column에서 bare html_generator.py
token 제거. §2.9 J5 (L979–981) SoT가 두 행 모두 부정확으로 lock —
A-3는 html_generator에 selenium import 부재, A-4는 slide-base 호출
부재 (area HTML만 반환). IMP-22 (2ace54b5590ef2noreply@anthropic.com >
2026-05-20 14:39:11 +09:00
2ace54bce1
docs( #22 ): PHASE-Q-AUDIT §1 lens A-3/A-4 renderer 칸 정정
...
A-3 main = slide_measurer.capture_slide_screenshot; renderer.py 는 간접
(render-path 자료); "selenium 캡처 흔적 추정" 제거.
A-4 = renderer.py (legacy slide-base.html 호출 지점 보유,
embedded/standalone CSS 분기 미구현) 추가. html_generator.py token
양 행 보존 (IMP-23 boundary).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-20 14:02:53 +09:00
5590ef20b5
docs( #21 ): PHASE-Q-AUDIT §1 lens B-1/B-2 content_editor.py 오기재 제거
...
§1 lens 표 B-1 행(L104) "Phase Q 후보 파일" 칸 — content_editor.py 토큰 제거 (pipeline.py 보존).
§1 lens 표 B-2 행(L105) "Phase Q 후보 파일" 칸 — content_editor.py 토큰 제거 (글벗 fmt_slide.py html_to_slide_mdx 보존).
§2 모듈 리스트 L125 content_editor.py 항목의 (B-1, B-2) axis annotation 제거 (모듈 자체는 §2.6 audit 대상으로 유지).
근거: content_editor.py = slot-fill / Kei editor only (fill_content :73, fill_candidates :335). B-1 (zone-section override) = pipeline.py / composition planner 영역. B-2 (HTML→MDX 역변환) = 글벗 fmt_slide.py html_to_slide_mdx (외부 sibling). 로컬 src/phase_z2_verification_utils.py 는 B-2 검증 utility 만 (extract_text_from_html L64-73 / normalize_for_comparison L89-104 / strip_meta_lines L147-166). §2.6 G2 self-catch (L642-643, L679) 와 정합.
Scope: docs-only, 1 file changed, 3 insertions(+) / 3 deletions(-). src/ templates/ frontend 무변경. IMP-25 (pipeline.py B-1 precision 축) / 외부 fmt_slide.py reference 축 / content_editor archive verdict (§2.6) / PHASE-Q-INSIGHT-TO-22STEP-MAP.md:136 catch record 모두 보존.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-20 13:23:32 +09:00
134f52d3d3
feat( #58 ): L3 dormant trigger guard -- DORMANT-TRIGGERS.yaml + checker + orchestrator hook
...
P5-1 docs/architecture/DORMANT-TRIGGERS.yaml -- 5 entries (IMP-16/17/18/19 active + IMP-20 followup-linked #55 ).
P5-2 scripts/check_dormant_triggers.py -- standalone, reads registry, scans tree + diff, writes .orchestrator/dormant_alerts.json, exit 0 always.
P5-3 orchestrator.py -- _check_dormant_triggers() helper + Stage 4->5 informational alert branch (skips audit-only, never blocks).
P5-4 tests/orchestrator_unit/test_dormant_triggers.py -- 30 cases (yaml schema, registry contents, checker matching, false-positive guards, manual-evidence skip, orchestrator branch, audit bypass, governance ref).
P5-5 PROJECT-INTENT-AND-GOVERNANCE.md -- single anti-patterns row referencing the L3 registry as binding contract surface.
Tests: pytest -q tests = 337 passed (baseline 307 + 30 new).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-20 09:43:14 +09:00
