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
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
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
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
7d5639ad72
feat(IMP-13): A-3 — build-time frame preview generator (capture_slide_screenshot salvage)
...
scripts/generate_frame_previews.py iterates figma_to_html_agent/blocks/{frame_id}/index.html,
renders preview.png via Selenium headless (capture_slide_screenshot pattern reuse), and writes
_preview_manifest.json (schema v1) with idempotent stale-detect (mtime+sha256). Build-time only
— no runtime pipeline integration, no AI calls, no MDX/Jinja regen. Stage 2 baseline (commit
56619a0noreply@anthropic.com >
2026-05-18 06:25:05 +09:00
73a98b8ad1
IMP-04 F17 schema correction — paired_rows_4x2 + pill alternation + source-faithful theme
...
source = 8 atomic issues (4 paired rows × 2 cells per texts.md), 이전 strict-4
가 source 의 절반 누락. round 55~73 review-loop 의 calibration frame.
- contract : source_shape=top_bullets / layout_variant=paired_rows_4x2_alternating_pills
/ strict 8 (no pad/truncate) / role_order row_{1..4}_{left,right} / visual_hints
pill_positions + row_gap_after / builder paired_rows_4x2_slots
- builder : new _build_paired_rows_4x2_slots — 2-axis (row × side) deterministic
index mapping, strict 8 raises before render, quadrant_item parser 재사용
- partial : 4-row × 2-cell flex, pill alternation (row 1/3 top, row 2/4 bottom
via column-reverse), row 2-3 visual gap, source-faithful color (rgb(204,82,0)
→rgb(136,55,0) title + #60A451 row border + rgba(250,237,203,0.15) bg + #0c271e
body + 2px dashed #60A451 cell 분할선), pill = CSS approximation (asset crop
variant single-pass 비용 高 → fallback per Codex round 62/68 scope cap, pill
shape + alternation + green/cream/brown theme 보존), no row headers (source
부재, inference 금지)
- fixture : flat 8 top-bullet (texts.md 8 issues 그대로)
- smoke + R3 : PASS (11/11 self-check, 5535 chars partial, 8 units rendered,
pill alternation 정합, row 2-3 gap, no invented row headers)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-13 15:13:46 +09:00
5c27c492ba
feat(catalog): activate bim_current_problems_paired (IMP-04 Track A 8/16)
...
V4 signal = restructure 1 (4 MDX sample). Catalog-completeness activation —
Codex round 51 guardrail (per-frame source-evidence check, not blind F16
reuse). Source confirmed as 2x2 paired-rows BIM problem layout, distinct
from F16's quadrant-4 framing.
3-layer architecture (matrix §0) :
- V4 = matching authority — restructure tier signal; runtime activation
prepares Phase Z to assemble this frame when V4 ranks it.
- figma_to_html (1171281194) = source/evidence — 4 BIM problem cards in
2x2 grid (개념 부재 / 잘못된 접근방식 / 방향성 상실 / 전제조건 오류).
- Phase Z = runtime — adds catalog + partial + smoke fixture.
Builder reuse :
- `quadrant_flat_slots` reused with pad_to=4 + `issue_{n}_label/body` keys.
- `quadrant_item` parser reused.
- F16 quadrant pattern reused, but the partial is a 2-row × 2-column
problem-theme grid (red/orange/amber/deep-amber), not the F16 TL/TR/BL/BR
quadrant visual. Source-evidence-driven decision per Codex round 51 §10.
3 file changes :
1. templates/phase_z2/families/bim_current_problems_paired.html
- 2x2 CSS grid with per-issue problem theme.
- PROMOTED CSS : per-cell warning gradient (red/orange/amber/deep-amber),
title gradient (zone-title family), "!" bullet markers in per-cell color.
- NOT PROMOTED : Figma source banner / numbered badges / texture —
figma_to_html source evidence preserved for future fidelity review.
- ADAPTED : Figma absolute positioning → CSS grid 2x2, token-fixed
typography.
2. templates/phase_z2/catalog/frame_contracts.yaml — F17 contract appended
- frame_id=1171281194, family=cards, source_shape=top_bullets, strict 4,
role_order=[issue_1..issue_4].
- visual_hints.min_height_px = 350 (F16/F14 class — 2-row × 2-col density).
- accepted_content_types = [text_block].
- payload.builder = quadrant_flat_slots reuse with issue_{n}_* keys.
3. scripts/smoke_frame_render.py — bundled fixture for F17.
Verification :
- python scripts/smoke_frame_render.py --self-check : PASS 11/11 (F17 at
3856 chars CSS-only)
- python scripts/smoke_frame_render.py bim_current_problems_paired
--render-to data/runs/imp04_f17_visual : PASS, 0 raster refs
- python run_mdx03_pipeline.py --phase-z2 --run-id imp04_f17_regression :
PASS (MDX 03 V4 rank-1 unchanged; F17 not in MDX 03 V4 selection so
this is non-impacting regression check)
scope-lock honored : V4 logic / mapper / production render / Phase R' /
AI/Kei / 10 existing partials all unchanged.
4-class status :
- class 1 readiness : ✅
- class 2 content-fit : watch — paragraph-heavy source (each issue body
is multi-line Korean text). Compact 2x2 cell may need wrap. max-content
fit checked via R3 artifact.
- class 3 : N/A
- class 4 : N/A
Refs Gitea #4 (IMP-04 Track A frame 8 — V4 RS tier, source-evidence-confirmed)
2026-05-13 13:39:30 +09:00
735e58420e
feat(catalog): activate sw_reality_three_emphasis (IMP-04 Track A 7/16)
...
Catalog-completeness activation — NOT V4 endorsement (Codex round 49 §1+§2
guardrail). V4 signal = 0 across the current 4 MDX sample. Phase Z adds
runtime availability so V4 can route to this frame for future MDX content.
V4 remains the matching authority.
3-layer architecture (matrix §0) :
- V4 = matching authority — current sample evidence does NOT rank this frame
for the 4 MDX. activation must not be read as V4-driven promotion.
- figma_to_html (1171281209) = source/evidence — 3-emphasis problem cards
(토목 전문성 부족 / 비효율성 / 실무 적용 불가). full A+T+I+F+S.
- Phase Z = runtime — catalog + partial + smoke fixture added.
Builder reuse (per Codex round 49 §9 — single-frame, no batch yet) :
- `quadrant_flat_slots` reused with pad_to=3 + `emphasis_{n}_label/body`.
- `quadrant_item` parser reused.
- mapper.py unchanged.
3 file changes :
1. templates/phase_z2/families/sw_reality_three_emphasis.html
- 3-column grid with problem-emphasis theme (red/orange/amber per
column — frame intent is 약점/한계 진단).
- PROMOTED CSS : per-emphasis warning color, title gradient, bullet
markers in per-emphasis color, "!" bullet for warning semantic.
- NOT PROMOTED : Figma source decoration / banner / texture.
figma_to_html source evidence preserved.
- ADAPTED : Figma absolute → flex 3-column grid, token-fixed typography.
2. templates/phase_z2/catalog/frame_contracts.yaml — F28 contract appended
- frame_id=1171281209, family=cards, source_shape=top_bullets, strict 3,
role_order=[emphasis_1/2/3].
- visual_hints.min_height_px = 320.
- accepted_content_types = [text_block].
- payload.builder = quadrant_flat_slots (reuse) with emphasis_{n}_* keys.
3. scripts/smoke_frame_render.py — bundled fixture for F28.
Verification :
- python scripts/smoke_frame_render.py --self-check : PASS 10/10 (F28 at
3363 chars CSS-only)
- python scripts/smoke_frame_render.py sw_reality_three_emphasis
--render-to data/runs/imp04_f28_visual : PASS, 0 raster refs
- python run_mdx03_pipeline.py --phase-z2 --run-id imp04_f28_regression :
PASS
scope-lock honored : V4 logic / mapper / production render / Phase R' /
AI/Kei / existing 9 partials all unchanged.
4-class status :
- class 1 readiness : ✅
- class 2 content-fit : watch (problem description text may be longer than
3-card body budget)
- class 3 : N/A (V4-zero, catalog-completeness — not a mapping concern)
- class 4 : N/A
Refs Gitea #4 (IMP-04 Track A frame 7 — V4-zero catalog completeness)
2026-05-13 13:27:23 +09:00
46ff611a54
feat(catalog): activate info_management_what_how_when (IMP-04 Track A 6/16)
...
Catalog-completeness activation — NOT V4 endorsement (Codex round 47 §7
guardrail). V4 signal = 0 across the current 4 MDX evidence sample, but
the 32-frame all-in scope means runtime backend must accept this frame
when V4 ranks it for any future MDX outside the current sample.
V4 remains the matching authority. Phase Z adds runtime availability only.
3-layer architecture (matrix §0) :
- V4 = matching authority — current evidence does not rank this frame for
the sampled 4 MDX, so this activation must not be read as a V4-driven
promotion. It is purely catalog completeness so V4 can route to this
frame when warranted in future MDX content.
- figma_to_html (1171281179) = source/evidence — analysis/texts/index/flat/
assets all present, source intent is the What/How/When 3-section
framework.
- Phase Z = runtime — this commit adds catalog contract + partial + smoke
fixture. Builder reuse, no new builder/parser.
Builder reuse (Codex round 47 secondary criterion under exhausted V4 tier) :
- `quadrant_flat_slots` reused (F16/F11/F20 pattern) with pad_to=3 +
`section_{n}_label/body` key patterns.
- `quadrant_item` parser reused.
- mapper.py unchanged.
3 file changes :
1. templates/phase_z2/families/info_management_what_how_when.html
- 3-column grid (What / How / When), each with header + body bullets.
- PROMOTED CSS : per-section accent color (blue #2563eb What / orange
#ea580c How / green #16a34a When), title gradient (#000 → #883700
zone-title family), bullet markers in per-section accent color.
- NOT PROMOTED (P1 case-by-case + preservation guardrail) : Figma
상단 banner / icon / 배경 텍스처 / 장식. figma_to_html source
evidence preserved for future fidelity review.
- ADAPTED : Figma 50px+ → token-fixed, absolute positioning → flex grid.
2. templates/phase_z2/catalog/frame_contracts.yaml — F8 contract appended
- frame_id=1171281179, family=cards, source_shape=top_bullets, strict 3,
role_order=[section_1 (What), section_2 (How), section_3 (When)].
- visual_hints.min_height_px = 320 (F11/F20 class — 3-column header+body).
- accepted_content_types = [text_block].
- 3 sub_zones (section_1/2/3 main_text).
- payload.builder = quadrant_flat_slots (reuse) with section_{n}_* keys.
3. scripts/smoke_frame_render.py — bundled fixture for F8 self-check.
Verification :
- python scripts/smoke_frame_render.py --self-check : PASS 9/9 (F8 added at
3577 chars CSS-only)
- python scripts/smoke_frame_render.py info_management_what_how_when
--render-to data/runs/imp04_f8_visual : PASS, R3 artifact, 0 raster refs
- python run_mdx03_pipeline.py --phase-z2 --run-id imp04_f8_regression :
PASS (MDX 03 V4 rank-1 unchanged; F8 V4 signal = 0 across 4 MDX sample,
so not selected for MDX 03 either — purely catalog completeness)
scope-lock honored (3-layer + 4-class) :
- V4 logic / V4 evidence yaml : unchanged
- Existing PAYLOAD_BUILDERS / ITEM_PARSERS : unchanged (reuse only)
- Existing 8 partials : unchanged
- Composition planner / production render / Phase R' / AI/Kei : unchanged
4-class status :
- class 1 readiness : ✅ contract + builder reuse + partial + smoke + R3
- class 2 content-fit : watch — header single-line, body 3-5 bullets per
column. 6+ bullets per column may overflow.
- class 3 / 4 : N/A
Refs Gitea #4 (IMP-04 Track A frame 6 — V4-zero catalog completeness)
2026-05-13 13:19:03 +09:00
bc58102b66
feat(catalog): activate dx_sw_necessity_three_perspectives (IMP-04 Track A 5/16)
...
Reason : V4 LE=2 (03-1 + 01-1) + RS=1 — V4 LE tier strongest remaining
after F12/F11/F18 UAI tier. Track A frame 5 per Codex round 45 V4-priority
acceptance.
3-layer architecture (matrix §0) :
- V4 = matching authority — V4 ranked this frame light_edit for 03-1 (DX 시행
필수 요건) and 01-1 (용어 정의).
- figma_to_html (1171281198) = source/evidence — 386-line index.html + assets/.
- Phase Z = runtime — this commit adds catalog + partial + smoke fixture.
Builder reuse (no new builder/parser introduced) :
- Reuses existing `quadrant_flat_slots` builder (F16/F11 pattern) with
pad_to=3 + `perspective_{n}_label` / `perspective_{n}_body` keys.
- Reuses existing `quadrant_item` parser.
- Same flat-keyed label+body grammar as F11; only N=3 + key names differ.
- mapper.py unchanged — secondary builder reuse per Codex round 45.
3 file changes :
1. templates/phase_z2/families/dx_sw_necessity_three_perspectives.html
- Adapted from figma_to_html_agent/blocks/1171281198/index.html.
- 3-column grid (BIM 전면설계 / 디지털 전환 S/W / 고부가가치 산업전환).
- PROMOTED CSS : header bar dark green (#296B55 → #123328 Figma green
family), header text white bold, title gradient (#000 → #883700
F13/F14/F12/F11/F18 zone-title family), card border + bullet markers
(green family).
- NOT PROMOTED (P1 case-by-case + preservation guardrail per Codex
round 37) : 상단 dark green banner ("디지털 전환(DX)은 S/W가 필수다"
visual block), 좌측 DX circular area (multi-image + center text +
decor — main rhetorical anchor but cannot fit compact zone),
hanmaek/한자/배경 텍스처. figma_to_html source evidence preserved.
- ADAPTED : Figma 90/65/40 px → token-fixed, 1280×426 absolute +
zoom → Phase Z 3-column grid, 3 perspective cards → flex column.
2. templates/phase_z2/catalog/frame_contracts.yaml
- F20 contract appended.
- frame_id=1171281198, family=cards, source_shape=top_bullets, strict 3,
role_order=[perspective_1, perspective_2, perspective_3].
- visual_hints.min_height_px = 320 (3 col × header 30 + body bullets ~75
+ title 30 + padding 30 = ~195 + 125 buffer for label/5+ bullet
overflow; F12/F11/F18 class).
- accepted_content_types = [text_block].
- 3 sub_zones (perspective_1/2/3 main_text).
- payload.builder = quadrant_flat_slots (reuse) with perspective_{n}_*
key patterns.
3. scripts/smoke_frame_render.py — bundled fixture for F20 self-check.
Verification :
- python scripts/smoke_frame_render.py --self-check : PASS 8/8 (F20 added
at 3160 chars CSS-only)
- python scripts/smoke_frame_render.py dx_sw_necessity_three_perspectives
--render-to data/runs/imp04_f20_visual : PASS, R3 artifact, 0 raster
refs (CSS-only)
- python run_mdx03_pipeline.py --phase-z2 --run-id imp04_f20_regression :
PASS (MDX 03 V4 rank-1 still F13/F29; F20 light_edit candidate for 03-1
but F13 was use_as_is at higher rank, so F20 not selected here)
scope-lock honored (3-layer + 4-class) :
- V4 logic / V4 evidence yaml : unchanged
- Existing PAYLOAD_BUILDERS (5 builders) : unchanged. No new builder added.
- Existing ITEM_PARSERS (3 parsers) : unchanged. No new parser added.
- Existing 7 partials : unchanged.
- Composition planner / production render / Phase R' / AI/Kei : unchanged.
4-class status :
- class 1 readiness : ✅ contract + builder reuse + partial + smoke fixture
+ R3 artifact aligned.
- class 2 content-fit : watch — header single-line, body 3-5 bullets per
column. 6+ bullets per column may overflow.
- class 3/4 : N/A.
Codex review remains useful (per scope-lock §5 "shared catalog/builder
logic" category — quadrant_flat_slots is now reused by F16/F11/F20). New
builder/parser path is NOT this commit.
Refs Gitea #4 (IMP-04 Track A frame 5 — V4 LE tier, builder reuse)
2026-05-13 12:28:49 +09:00
c7b0f5bde1
feat(catalog): activate bim_dx_comparison_table (IMP-04 Track A 4/16)
...
Reason : V4 UAI=1 (01-2 "용어간 상호관계") — UAI tier strongest after F12/F11.
Track A frame 4 per Codex round 41 V4-priority acceptance.
3-layer architecture (matrix §0) :
- V4 = matching authority — V4 ranked this frame use_as_is for 01-2.
- figma_to_html (1171281195) = source/evidence — analysis/texts/index.html/
flat/assets all present.
- Phase Z = runtime orchestration — adds catalog + new builder + new parser +
new partial + smoke fixture.
NEW builder + NEW parser (Codex round 41 mandatory review path) :
1. src/phase_z2_mapper.py — NEW `compare_row_2col_item` parser in ITEM_PARSERS
- input : (top_line, nested_lines)
- output : {label, col_a, col_b}
- label = bold from top_line
- col_a / col_b = first 2 nested bullets, optional prefix stripping ("BIM:"/
"DX:" or similar ≤8-char tag with colon)
- inline emphasis preserved as <strong>
2. src/phase_z2_mapper.py — NEW `compare_table_2col` PAYLOAD_BUILDERS entry
- payload : title + col_a_label + col_b_label + rows[]
- builder_options : item_parser, col_a/b_label_default, max_rows (default 999)
- max_rows truncation tracked via _truncated_count
3. templates/phase_z2/families/bim_dx_comparison_table.html — NEW partial
- 3-column grid (category / col_a / col_b) with header row + N data rows
- PROMOTED CSS : title gradient (#000 → #883700 , zone-title family), header
brown bg (rgba(50,31,9,0.85-0.95)), zebra striping, brown family bullet
accent, subtle border (#A5BBB4 F11 family).
- NOT PROMOTED (P1 case-by-case + preservation guardrail) : Figma column
header raster icons, color emphasis variants, hanja deco. figma_to_html
source evidence remains preserved.
- ADAPTED : Figma absolute positioning + zoom → Phase Z flex/grid 3-col
table, typography → token-fixed, row heights auto content-fit.
4. templates/phase_z2/catalog/frame_contracts.yaml — F18 contract appended
- frame_id=1171281195, family=table, source_shape=top_bullets, strict 2
(2 columns), role_order=[col_a, col_b].
- visual_hints.min_height_px = 350 (title 30 + header 30 + 6 rows×35 +
padding 30 = 300 + 50 buffer; F14-class).
- accepted_content_types = [text_block].
- sub_zones : col_a_header / col_b_header (strict 1 each) + rows (min 1,
max 12 category rows).
5. scripts/smoke_frame_render.py — bundled fixture for F18 self-check (6
category rows : 범위 / S/W / 프로세스 / 성과물 / 활용 / 수행개념).
Verification :
- python -m py_compile src/phase_z2_mapper.py scripts/smoke_frame_render.py
: PASS
- python scripts/smoke_frame_render.py --self-check : PASS 7/7 (F18 added
at 4211 chars CSS-only)
- python scripts/smoke_frame_render.py bim_dx_comparison_table --render-to
data/runs/imp04_f18_visual : PASS, R3 artifact, 0 raster refs (CSS-only)
- python run_mdx03_pipeline.py --phase-z2 --run-id imp04_f18_regression :
PASS (MDX 03 V4 rank-1 still F13/F29; F18 only routes 01-2 per V4)
scope-lock honored (3-layer + 4-class) :
- V4 logic / V4 evidence yaml : unchanged
- Existing PAYLOAD_BUILDERS (4 builders) : unchanged. compare_table_2col added
as NEW entry.
- Existing ITEM_PARSERS (2 parsers) : unchanged. compare_row_2col_item added
as NEW entry.
- Existing 6 partials : unchanged.
- Composition planner / production render / Phase R' / AI/Kei : unchanged.
4-class status :
- class 1 readiness : ✅ contract + new builder + new parser + partial +
smoke fixture + R3 artifact aligned.
- class 2 content-fit : watch — cell content single-line; long Korean
sentences may wrap. Row height auto handles wrap; max_rows=12 limit
protects vertical overflow.
- class 3/4 : N/A.
Codex review mandatory per scope-lock §5 (new builder pattern first
introduction : compare_table_2col).
Refs Gitea #4 (IMP-04 Track A frame 4 — V4 UAI tier, NEW builder)
2026-05-13 12:13:11 +09:00
a4fdc7ad89
feat(catalog): activate construction_bim_three_usage (IMP-04 Track A 3/16)
...
Reason : V4 UAI=1 (01-1) + RS=1 — UAI tier strongest after F12. Track A
frame 3 per Codex round 39 V4-priority acceptance.
3-layer architecture context (matrix §0) :
- V4 = matching authority — V4 ranked this frame as use_as_is for 01-1
("용어 정의") and restructure=1 for other sections.
- figma_to_html (1171281182) = rich source/evidence — 401-line index.html
+ 4 PNG assets, full A+T+I+F+S resource available.
- Phase Z = runtime orchestration — this commit adds catalog + partial +
smoke fixture for the runtime projection.
Builder reuse (Codex round 39 secondary criterion) :
- Reuses existing `quadrant_flat_slots` (F16) with pad_to=3 +
label_key_pattern="category_{n}_label" + body_key_pattern="category_{n}_body".
- Same parser `quadrant_item` reused — no new builder or parser added.
- F16 / F11 share flat-keyed label+body grammar; only N differs (4 vs 3).
3 file changes :
1. templates/phase_z2/families/construction_bim_three_usage.html
- Adapted from figma_to_html_agent/blocks/1171281182/index.html.
- 3 horizontal stacked rows (brown label box + white card with green
border, per Figma L82-103). Compact Phase Z zone fit.
- PROMOTED CSS : brown label box (rgba(50,31,9,0.8) + 둥글기 + drop-
shadow), white card with green border (#A5BBB4), title gradient
(#000 → #883700 F13/F14/F12 family), bullet markers in brown family.
- NOT PROMOTED (P1 case-by-case + preservation guardrail per Codex
round 37 / matrix §4.1 Fix 7) : title-icon (50×50 small), card_decor
× 3 (decorative side images), bottom BIM 모델/공사 Process 다이어그램.
Figma source evidence preserved in figma_to_html_agent/; not promoted
to Phase Z compact projection.
- ADAPTED : Figma 50/45/38/30 px → token-fixed, 1248×1066 absolute +
zoom 0.67542 → Phase Z flex column 3-row stack, 335×124 label box
→ 110×~50 compact (zone fit).
2. templates/phase_z2/catalog/frame_contracts.yaml
- F11 contract appended after F12.
- frame_id=1171281182, family=cards, source_shape=top_bullets, strict
cardinality 3, role_order=[category_1, category_2, category_3].
- visual_hints.min_height_px = 320 (title 30 + 3×70 row + gaps 12 +
padding 30 = 282 + 38 safety buffer).
- accepted_content_types = [text_block] only.
- 3 sub_zones (category_1/2/3 main_text).
- payload.builder = quadrant_flat_slots (reuse) with category_{n}_*
key patterns.
3. scripts/smoke_frame_render.py
- Bundled fixture for F11 self-check.
Verification :
- python -m py_compile scripts/smoke_frame_render.py : PASS
- python scripts/smoke_frame_render.py --self-check : PASS 6/6 (F11
added at 3014 chars, compact CSS-only)
- python scripts/smoke_frame_render.py construction_bim_three_usage
--render-to data/runs/imp04_f11_visual : PASS, R3 artifact written,
0 raster refs (CSS-only), copy_assets ran (4 PNGs in assets/ dir
for future fidelity-review if needed)
- python run_mdx03_pipeline.py --phase-z2 --run-id imp04_f11_regression
: PASS (MDX 03 V4 rank-1 still F13/F29, F11 not triggered for MDX 03 —
F11 only routes 01-1 per V4 evidence)
scope-lock honored (3-layer + 4-class) :
- V4 logic / V4 evidence yaml : unchanged
- Existing PAYLOAD_BUILDERS 4 builders (incl. F12's cycle_intersect_3) :
unchanged. F11 reuses quadrant_flat_slots (secondary criterion).
- Existing ITEM_PARSERS : unchanged. F11 reuses quadrant_item.
- Existing partials (F13/F29/F16/F14/F12) : unchanged.
- Composition planner / production render path / Phase R' / AI/Kei :
unchanged.
4-class status :
- class 1 adapter/runtime readiness : ✅ contract + builder + partial +
smoke fixture + R3 artifact aligned.
- class 2 content-fit : watch — 3 rows × ~70 px height. If MDX body has
4+ bullets per category, may overflow. realistic use case (01-1) shows
1-2 bullets per category per analysis.md, so within budget.
- class 3 / 4 : N/A.
Refs Gitea #4 (IMP-04 Track A frame 3 — V4 UAI tier, builder reuse)
2026-05-13 12:05:05 +09:00
766fa4639d
fix(IMP-04): F12 F1 follow-ups — intersection optional + smoke title generic
...
Same-frame F1 follow-up commit per Codex round 37 (#15503 ). Matrix §4.1
Fix 7 4-class: F1 = small same-frame fix + Codex re-review (not F2/F3,
no Track A pause needed).
Two fixes (both Codex-caught) :
1. F1-a — `intersection` sub_zone contract semantics
- Previous (c67609c#4 (IMP-04 Track A frame 2 — F1 follow-up per Codex round 37)
2026-05-13 11:57:45 +09:00
c67609c083
feat(catalog): activate construction_goals_three_circle_intersection (IMP-04 Track A 2/16)
...
Reason : V4 strongest UAI tier candidate (use_as_is=1 for 02-1, light_edit=1
for 01-1, restructure=1). Track A frame 2 per Codex rounds 30/33/35 V4-
priority rule. F14 clean pass completed at 834ed39; this is the next
Track A activation.
3-layer architecture context (matrix §0) :
- V4 = matching authority — V4 ranked this frame as use_as_is for the
"DX의 궁극적 목표" section (02-1) and light_edit for "용어 정의" (01-1).
- figma_to_html (1171281189) = rich source/evidence — 510-line index.html
base, full analysis/flat/texts/assets present (A+T+I+F+S).
- Phase Z = runtime orchestration — this commit adds the runtime contract,
builder, partial, and fixture so the V4 candidate can be assembled.
New runtime additions :
1. src/phase_z2_mapper.py — new `cycle_intersect_3` PAYLOAD_BUILDERS entry
- Reuses existing `quadrant_item` ITEM_PARSERS (label only, body
ignored) — F16 parser reused, no new parser.
- Produces flat keys : circle_1_label / circle_2_label / circle_3_label
+ intersection text (optional) — distinct from F16's quadrant_N_body
structure since this frame's 3 main circles use labels only.
- pad_to=3, truncate_at=3, configurable via builder_options.
2. templates/phase_z2/families/construction_goals_three_circle_intersection.html
- Adapted from figma_to_html_agent/blocks/1171281189/index.html.
- Slot mapping : title + 3 circle labels + optional intersection text.
- PROMOTED CSS : 3 circle gradients (safety #BC652B/#A24200, productivity
#897445/#3E3523, trust #296B55/#123328) + outer multiply blend +
title gradient (#000 → #883700 , F13/F14 zone-title family) + main
label typography (white text + shadow).
- NOT PROMOTED (P1 case-by-case, compact zone fit) : 6 accent hanja
circles (安/質/速/利/通/信), 6 side labels (안전성 제고 etc.), 3
decoration rects, 3 arc images, bg-texture multiply image. These
are Figma-side decorative content not in MDX and would clutter a
Phase Z zone of ~320 px.
- ADAPTED : Figma 70/50/40 px → token-fixed font sizes, 350×350
absolute-positioned overlapping circles → 110×110 flex row (cycle
intent expressed via intersection text instead of geometric overlap).
3. templates/phase_z2/catalog/frame_contracts.yaml — append F12 contract
- template_id, frame_id 1171281189, family=diagram, source_shape=
top_bullets, strict cardinality 3, role_order [safety, productivity,
trust].
- visual_hints.min_height_px = 320, derived from 3 circle row 80 +
title 30 + label area 60 + intersection 30 + padding 40 = 240
+ 80 safety buffer (lighter than F14's 350 since CSS-only).
- accepted_content_types = [text_block] only.
- 4 sub_zones declared (circle_1/2/3 main_text + intersection emphasis).
4. scripts/smoke_frame_render.py — add bundled fixture for F12 self-check.
Verification :
- python -m py_compile src/phase_z2_mapper.py scripts/smoke_frame_render.py
: PASS
- python scripts/smoke_frame_render.py --self-check : PASS 5/5 (F12 added
at 3691 chars CSS-only)
- python scripts/smoke_frame_render.py construction_goals_three_circle_intersection
--render-to data/runs/imp04_f12_visual : PASS, R3 artifact written. 0
raster refs (CSS-only partial); copy_assets ran successfully and
produced data/runs/imp04_f12_visual/assets/construction_goals_three_circle_intersection/
with the frame's 4 PNG files (unused since partial is CSS-only — assets
remain available for future raster promotion if visual inspection
flags fidelity loss).
- python run_mdx03_pipeline.py --phase-z2 --run-id imp04_f12_regression
: PASS (MDX 03 V4 rank-1 still F13/F29, F12 not selected — F12 only
triggered by 02-1 / 01-1 sections per V4 evidence)
scope-lock honored : V4 logic / V4 evidence / mapper existing builders /
composition planner / Phase R' / pipeline production render path / AI/Kei
all unchanged. New builder added without modifying existing 3 (mixed
strategy per scope-lock §4).
Calibration status (matrix §4.1 Fix 7 4-class) :
- class 1 adapter readiness : new builder registered, partial loadable,
contract valid, smoke passing.
- class 2 content-fit : compact 110×110 circles + label, watch for label
overflow if MDX bullets exceed ~12 chars.
- class 3/4 mapping/routing : not applicable for this commit.
- Codex review mandatory per scope-lock §5 (new builder pattern
cycle_intersect_3 first introduction).
Refs Gitea #4 (IMP-04 Track A frame 2 — V4 strongest UAI tier)
2026-05-13 11:50:44 +09:00
834ed3946d
test(IMP-04): add F14 render artifact check and fix min-height note
...
3rd commit in F14 series (calibration point clean pass). Closes the two
Codex round 26 (#15435 ) blockers:
1. min_height_px self-contradiction
2. F14 actual rendered visual artifact absent
Per Codex round 28 (#15447 ) agreement (M1 + --render-to extension) and
Claude round 27 (#15438 ) fix path :
Changes :
1. templates/phase_z2/catalog/frame_contracts.yaml — min_height_px 320 → 350.
Comment now self-consistent : 70 (badge raster) + 210 (bullet body) +
36 (photo strip) + 30 (padding) = 346 sum + 4 safety buffer = 350.
F14 is now F29-class (345) per raster-promoted content density.
2. scripts/smoke_frame_render.py — add `--render-to DIR` dev mode (R3
acceptance gate). Behavior :
- StrictUndefined smoke render (unchanged)
- reuse production `copy_assets(template_id, run_dir)` so the runtime
asset delivery path is exercised (no logic duplication)
- wrap partial with minimal viewer HTML (Phase Z token vars + slide-
sized wrap, browser-openable)
- fail-fast if rendered HTML references a missing local asset (per
Codex round 28 §4 recommendation)
- save artifact to {DIR}/index.html with {DIR}/assets/{template_id}/*
- production render path (phase_z2_pipeline.render_slide) unchanged
- small regex fix : asset extraction now captures both `src="..."`
and `url("...")` references
F14 verification (3rd commit) :
- python -m py_compile scripts/smoke_frame_render.py : PASS
- python scripts/smoke_frame_render.py --self-check : PASS 4/4 (7446 chars
for persona unchanged from a1c06b7a1c06b7#4 (IMP-04 Track A — F14 3rd commit, clean pass gate)
2026-05-13 11:07:37 +09:00
556b4486ae
feat(catalog): activate three_persona_benefits frame (IMP-04 #4 / 1 of 7)
...
Reason : V4 use_as_is=1 (frame_number=14, frame_id=1171281191).
Pattern : cards-3col-persona — 발주자/시공자/설계자 3 주체 각 benefit.
- Append `three_persona_benefits` contract to frame_contracts.yaml after
the existing F13/F29/F16 entries (Codex Catch 1/4: YAML order = trace
selection surface)
- Reuse existing builder primitives: items_with_role + quadrant_item
parser. No new entry in PAYLOAD_BUILDERS / ITEM_PARSERS.
Output dict shape: payload.personas = [{label, body, color_class}, ...]
- Add families/three_persona_benefits.html partial:
- Pure CSS (no Figma raster img tags) per memory rule
`feedback_blocks_must_be_css.md`
- PROMOTED colors per persona (#285b4a client / #445a2f constructor /
#743002 designer) from Figma TEXT layers
- NOT PROMOTED: col_bg_texture / overlay / 하단 사진 / 원형 뱃지 inner-outer
image — all replaced by CSS approximation (pill badge + colored
border + check-style text-line bullets)
- Token-fixed typography (zone-title / sub-title / caption / body)
- data-frame-id="1171281191" data-template-id attributes
- Add bundled smoke fixture for three_persona_benefits to
scripts/smoke_frame_render.py
- visual_hints.min_height_px = 280 (initial estimate between F13=230 and
F29=345 for 3-card text-heavy layout). Refine during batch full
pipeline if needed.
- accepted_content_types = [text_block] only (rich types not routed yet
per IMP-03 scope-lock).
Verification :
- isolated Jinja StrictUndefined smoke (scripts/smoke_frame_render.py
--self-check) : PASS=4/4 (existing 3 + new persona, 3889 chars)
- regression run on MDX 03 (env OFF + rich OFF) : PASS — MDX 03 V4
rank-1 still F13/F29 so the new entry does not affect existing flow
scope-lock 15 conditions all honored (no V4 / mapper / Phase R' / Step
6+ changes; per-frame 6-step gate complete; YAML order preserved).
Refs Gitea #4 (IMP-04 A-2 Catalog 확장)
2026-05-13 06:56:35 +09:00
2717a0a3a6
feat(infra): per-frame Jinja smoke harness with StrictUndefined (IMP-04 #4 )
...
- Add scripts/smoke_frame_render.py for IMP-04 scope-lock §11+§13:
isolated StrictUndefined Jinja partial render gate, separated from
production render_slide() permissive behavior
- CLI: --self-check runs every bundled fixture; positional template_id
takes payload via --payload <json> / stdin / fixture
- Bundled mock payloads for the 3 existing active frames match real
builder output shape; all 3 partials PASS self-check
- New frame activation gate (per-frame 6-step Step 5): partial must
pass smoke render with a complete mock that mirrors the builder's
output dict (optional fields supplied as empty/falsy so {% if %}
guards still work under StrictUndefined)
- Exit codes: 0=all pass, 1=at least one fail, 2=invalid input
Latent finding (out of IMP-04 scope, surfaced for the record):
bim_issues_quadrant_four partial references slot_payload.quadrant_N_headline,
but _build_quadrant_flat_slots() only produces quadrant_N_label and
quadrant_N_body. The headline div therefore never renders in production.
Either dead reference or a builder extension that was never landed —
reconcile in a follow-up axis, not in IMP-04 catalog expansion.
production render path (phase_z2_pipeline.render_slide) unchanged.
Refs Gitea #4 (IMP-04 A-2 Catalog 확장 — infra commit)
2026-05-13 06:48:19 +09:00
85c680f02a
docs + V4 catalog + samples + Phase Q legacy 보존
...
전체 26 files (20 추가 + 6 수정), 10507 insertions.
Phase Z 문서 :
- docs/architecture/PHASE-Z-CHANGE-LOG.md (신설) — axis-by-axis 의사결정 history
(newest-on-top). Step 7-A 부터 6 entry 박힘 + 2026-05-08 / 2026-05-08 #2
(compat 매트릭스 폐기 / 6-B 폐기 / F14 표현 정정 / label gate policy 분리).
- docs/architecture/PHASE-Z-PIPELINE-OVERVIEW.md (수정) — Step 5/6/9 Gap note
append (구조 무변, append-only). 6-B 폐기 사실 + Refinement F.
- docs/architecture/PHASE-Z-PIPELINE-STATUS-BOARD.md (수정) — snapshot date
2026-05-08 갱신. §3 핵심 missing item 5 (Step 5/6/9 boundary axis breakdown
+ 폐기 기록). §6 한 줄 갱신 — 다음 axis 후보 A~F.
Project root docs :
- PLAN.md / PROGRESS.md / README.md (수정) — 토큰 체계 / 폴더 구조 / 설계 문서 /
역할 분리 반영.
- IMPROVEMENT-REDESIGN.md (신설) — Phase Z 설계 핵심 문서.
- PROCESS_OVERVIEW.html (신설) — 파이프라인 개요 시각.
- docs/tasks/* (신설) — Phase Z task 문서.
V4 catalog (Phase Z runtime 필수 의존성) :
- tests/matching/v4_full32_result.yaml (신설, 4888 줄) — V4 매칭 결과 32 frame
× 10 MDX section. lookup_v4_match() / lookup_v4_candidates() 가 본 파일 read.
Phase Z runtime 이 *없으면 즉시 abort* — clone 후 즉시 동작 가능 보장.
Samples :
- samples/mdx_batch/04.mdx (신설) — MDX04 기본 sample.
- samples/mdx/04. DX 지연 요인.mdx (신설) — MDX04 원본.
Phase Q legacy 보존 (별 axis "Phase Q audit & salvage" 영역) :
- src/block_matcher_tfidf.py / catalog_blocks.py / frame_extractor.py /
pipeline_v2.py — Phase Q (옛 파이프라인) src 신규 untracked 파일들.
Phase Z runtime 와 의존성 0. Phase Q audit axis 에서 검토 예정.
- scripts/eval_block_matcher.py / fetch_all_frame_screenshots.py /
match_17_units_my_matcher.py / match_mdx_strict.py / match_mdx_to_frames_tfidf.py /
ocr_augment_texts.py / run_pipeline_v2.py / previews/ — Phase Q 작업 시
사용한 옛 script. 같이 보존.
- run_mdx03_pipeline.py (수정) — Phase Q 진입점 (no flag) + Phase Z 진입점
(--phase-z2 flag) 동시 wrapper. Phase Z 만 사용 시 `python -m
src.phase_z2_pipeline samples/mdx_batch/03.mdx <run_id>` 직접 호출.
비-scope :
- tests/matching/ (v4_full32_result.yaml 외 ~63MB) — V4 진화 history /
reports / DECK / ATTACH. Phase Q audit axis 에서 검토.
- tests/pipeline/ (~15MB) — pipeline data. Phase Q audit 영역.
- templates/catalog/blocks.yaml — 옛 block catalog. Phase Q audit.
- templates/phase_z2/frames/ — 옛 frame partial 위치. Phase Q audit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-05-08 09:47:58 +09:00
05703c8e72
WIP: hero-icon-cards_1 블록 + 오답노트 + figma 관련 파일
...
- hero-icon-cards_1.html: hero-icon-cards 변형 (icon → 소제목+불릿 계층)
- compare-detail-gradient.html: 하단 2열 비교 블록 (Figma Frame 4 기반)
- 오답노트.md: 절대 하지 말아야 하는 실수 목록
- figma_to_html.py: Figma→HTML 변환 스크립트
- static/figma-assets/: Figma export 이미지 (배지, 화살표)
- 주의: compare-detail-gradient CSS 폰트 크기가 임의 수정됨 — 원본 복원 필요
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-07 17:14:09 +09:00
f568e5c95d
이미지 마크다운 필터 추가:  패턴도 content_lines에서 제거
...
- block_assembler + assemble_stage2: 기존 [이미지:] 패턴에 ![markdown image 패턴 추가
- 02번 상단 "소통과 신뢰" 카드에 이미지 경로가 불릿으로 표시되던 문제 해결
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-07 07:45:32 +09:00
17e77e310f
Phase X-BX' XBX-1,3,5,6 완료: 유형 B 파이프라인 정상 동작
...
- XBX-1: normalizer 불릿 depth 보존 (D1/D2 마커) + 조립 로직 계층 반영
- XBX-3: 하단 구조 개선 — 하나의 큰 박스 안에 중제목 헤더 + 세로 구분선 2분할
- XBX-5: before→filled→after 파이프라인 연결 확인 (filled 2.2MB, 측정/재배분 정상)
- XBX-6: Type B에서 Sonnet 재구성 + renderer 스킵 — code_assembled 직접 사용
- final.html: 4,934 bytes → 2.2MB (Type B 정상 출력)
- Type A 코드 한 글자도 안 건드림
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-07 06:00:18 +09:00
d4eaec694c
유형 B 파이프라인 연결: block_assembler type B 조립 + zone 기반 전환 시작
...
- block_assembler: _assemble_slide_html_type_b 추가 (filled/after용 HTML 생성)
- fit_verifier: redistribute()가 ROLE_ZONE_MAP 대신 containers zone 사용
- renderer: render_slide_from_html()에 zone 기반 높이 탐색 추가
- pipeline: 팝업 HTML CSS를 콘텐츠 유형별(table/list/text) 분기
- run_from_stage1b: MDX 파일 하드코딩 제거 + layout_template 전달 추가
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-07 04:39:02 +09:00
4f0105926d
03번 MDX sections 매핑 수정: 상단 level=2 합침 + 하단 대목차 정확히 찾기
...
- _assemble_type_b: 상단에 해당하는 모든 level=2 section을 합침
(03번처럼 기술/사람/자연이 별도 section으로 분리된 경우 대응)
- 하단 대목차: level=3 바로 앞의 level=2 section으로 정확히 찾기
- 03번 결과: 상단 카드(기술/사람/자연) + 하단(과정혁신/결과변화) 정상
- 02번 영향 없음
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-06 13:33:05 +09:00
3719704d75
X' 핵심 수정: MDX sections에서 직접 텍스트 가져오기 + normalizer ### 지원
...
핵심 변경:
- mdx_normalizer: ### (h3) 소목차도 section으로 분리 (기존 ## 만)
- _assemble_type_b: Kei structured_text 대신 normalized.sections에서 직접 텍스트
- 대목차/소목차 계층 구조 그대로 반영
결과:
- 슬라이드 제목: 원본 MDX frontmatter 그대로
- 대목차: "DX 기반 Process 혁신에 따른 주체별 기대효과"
- 소목차 좌: "업무 수행 과정(Process)의 변화"
- 소목차 우: "DX 시행 주체별 기대효과" + 팝업 링크 + Kei 요약 표
- 캡션: normalized.images alt text
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-06 12:22:09 +09:00
6b17f448eb
Phase X'-6 추가: 본문 표 요약 프로세스 (미완성)
...
- pipeline.py: normalized.tables에서 본문 표 감지 → Kei 요약 요청
- assemble_stage2: _assemble_type_b 하단 우측에 table_summaries 표출
- 검증: 4열x3행 표 생성 확인
미해결:
- 들여쓰기 계층이 PNG와 다름 (대제목→소제목→본문 indent)
- 상단 컨테이너 내용 잘림
- 하단 우측: 표를 불릿으로 풀지 말고 팝업 링크 + 요약 표로
- [DX 시행 주체별 기대효과 바로가기 →] 팝업 처리
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-06 11:57:46 +09:00
56fd9fa71e
Phase X'-1~5 완료: 제목/들여쓰기/캡션/빈칸/카드 디자인
...
X'-1: 제목 원본 MDX frontmatter에서 가져오기 (Kei가 바꾸지 않음)
X'-2: 들여쓰기 계층 (소제목→불릿 indent 적용)
X'-3: 이미지 캡션 normalized.images alt text에서 추출
X'-4: 상단 컨테이너 justify-content:space-between
X'-5: 카드 디자인 다크 그라데이션 + 밝은 텍스트
X'-6 미완료: 본문 표(팝업 아닌)를 하단 우측에 Kei 요약 배치 → 다음 세션
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-06 11:40:53 +09:00
a8fe20e08e
Phase X-B 진행중: 유형 B 조립 + 텍스트 보존 강화 + 원본 MDX 복구
...
X-B-3~5 완료:
- space_allocator: build_containers_type_b() 추가
- assemble_stage2: _assemble_type_b() 추가 (소제목 카드형)
- pipeline.py: layout_template 분기 (A/B)
- pipeline_context: Analysis.layout_template 필드
- validators: 유형 B 검증 완화
텍스트 보존 강화:
- KEI_PROMPT: 제목 원본 그대로, 텍스트 재작성 금지
- KEI_STRUCTURED_TEXT_PROMPT: 소제목 유지, 원본 문장 그대로
원본 MDX 복구:
- samples/mdx_batch/02.mdx: 표 데이터 누락 수정 (원본에서 재복사)
미해결 (다음 세션):
- 들여쓰기: 대제목→중제목→소제목→본문 계층 구조
- 이미지 캡션: [그림 제목] 형식 (대괄호 포함)
- 상단 컨테이너: 빈칸 위로 붙이기
- 카드 디자인: 안전과품질/생산성향상/소통과신뢰 디자인 개선
- 제목: Kei가 원본 제목 바꾸는 문제 잔존
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-06 11:28:03 +09:00
c9677a69f8
V'-2/V'-4 수정: 표 행 수 계산 + footer 최소 높이 + 이미지 비율
...
- V'-2: 표 공간 계산에 V'-4(결론 위까지 채움) 높이 반영
→ Kei에게 정확한 행 수 전달 (1행 → 5행)
- V'-2: 이미지 높이를 실제 비율로 계산 (sub_layout 고정값 대신)
→ 200/2.73 = 73px (기존 172px → 공간 100px 확보)
- footer 최소 높이: design tokens 기반 동적 계산
→ weight 0.05일 때 26px → 53px 보장
- assemble_stage2: 이미지 높이도 실제 비율 반영
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-06 09:38:28 +09:00
1f7579cf64
Phase W + V' 완료: before→filled→after 파이프라인 + 조립 로직 수정
...
Phase W:
- weight 비율 초기 배정 (space_allocator header 높이 반영)
- block_assembler 공통 조립 함수 (filled/assembled 통합)
- filled → Selenium 측정 → context 저장
- sidebar overflow 확장 + body 재배분
- sub_layouts 사전 계산 (이미지 누락 해결)
Phase V':
- 팝업 링크 우측상단 배치 (인라인 → position:absolute)
- 표 내용 Kei 판단 (공란 크기 계산 → 행/열 산출 → Kei 요약)
- 출처 라벨 삭제 + 이미지 아래 캡션 배치
- after 공란 제거 (결론 바로 위까지 body/sidebar 채움)
추가:
- V-10 bold 키워드: 기계적 추출 → Kei 문맥 판단
- ** 마크다운 → <strong> 변환
- [이미지:] 마커 제거 (bold 변환 전 처리)
- grid-template-rows AFTER 크기 반영 (Sonnet final)
- assemble_stage2 CSS font-size override, white-space fix
- 하드코딩 전수 검토 완료
- 본심 여러 topic 텍스트 합침
Phase X 계획 문서 작성 (동적 역할 구조)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-06 05:00:52 +09:00
29f56187c0
Phase P~S 전체 작업물: 검증 스크립트, 블록 템플릿, 설계 문서, 코드 수정
...
포함 내용:
- Phase P/Q/R/S 설계 문서 (IMPROVEMENT-PHASE-*.md)
- 영역별 검증 스크립트 (scripts/verify_*.py, test_*.py)
- 블록 템플릿 추가 (cards, emphasis 변형)
- 코드 수정: block_search, content_editor, design_director, slide_measurer
- catalog.yaml 블록 목록 업데이트
- CLAUDE.md, PROGRESS.md, README.md 업데이트
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-03-31 08:38:06 +09:00
0e4b8c091c
Phase S: Claude HTML 직접 생성 + 독립 검증 시스템 도입
...
블록 선택 방식(Phase P/Q/R) 폐기 → Claude Sonnet이 영역별 HTML 직접 생성.
생성-검증 분리: content_verifier.py로 텍스트 보존/금지 콘텐츠/구조를 코드 검증.
주요 변경:
- src/html_generator.py: 4개 프롬프트 템플릿(BG/CORE/SIDEBAR/FOOTER) + 영역별 Claude 호출
- src/content_verifier.py: L1 텍스트 보존, L2 금지 콘텐츠, L3 구조 검증 + 재시도 루프
- src/html_validator.py: 보안 검증(script/iframe 제거)
- src/renderer.py: render_slide_from_html() 추가, area div overflow:hidden
- scripts/test_phase_s.py: generate_with_retry() 통합, step2b_verification 결과 저장
- 배경 라이트 디자인(#f8fafc), 개조식 어미 변환, 축약 금지 규칙
다음 과제: 폰트 위계(핵심14>본문12>배경10-12>첨부9-11) + 동적 컨테이너 계산
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-03-31 08:37:05 +09:00
b0bcffc0f6
Phase N+O: 컨테이너 기반 레이아웃 + Step B 제거 + 전면 정리
...
- Phase N: catalog 개선, fallback 전면 제거, Kei API 무한 재시도, topic_id 버그 수정
- Phase O: 컨테이너 스펙 계산(비중→px), 블록 스펙 확정, 렌더러 container div
- Step B(Sonnet) 제거: Kei(A-2)+코드로 대체. STEP_B_PROMPT/fallback/DOWNGRADE_MAP 삭제
- Selenium: container div 감지 추가
- catalog.yaml: ref_chars 구조 변환 + FAISS 재빌드
- 문서 전면 갱신: README, PROGRESS, IMPROVEMENT, Phase I~O md
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-03-27 15:20:51 +09:00
a01f7a7f8a
Phase G: Kei API 통신 정상화 — streaming 전환 + Sonnet fallback 제거
...
G-1: httpx non-streaming → streaming 전환 (3개 파일)
- client.post() → client.stream("POST") + response.aiter_lines()
- SSE 토큰을 실시간 수신 (30분+ 무응답 해소)
G-2: Sonnet fallback 완전 제거
- kei_client.py: classify_content()에서 _call_anthropic_direct() 호출 제거
- content_editor.py: fill_content()에서 Sonnet fallback 분기 제거
- Kei API만 사용. 실패 시 manual_classify() 또는 _apply_defaults() 안전망
G-3: _parse_json() 마크다운 제거 3파일 동기화
- content_editor.py, design_director.py에 kei_client.py와 동일한 전처리 추가
G-4: FAISS를 CPU로 전환 (GPU 메모리 경쟁 해소)
- block_search.py + build_block_index.py: device="cpu"
G-5: streaming 파서에 event:error 처리
- persona_agent 에러 시 무한 대기 방지. 즉시 중단.
G-6: content_editor.py None 가드
- Kei API 실패 시 _parse_json(None) TypeError 방지
G-7: "mode" → "mode_hint" 필드명 수정 (3개 파일)
- persona_agent의 실제 필드명에 맞춤
persona_agent 수정: 0건
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-03-26 01:26:03 +09:00
9bd9dad9ac
IMPROVEMENT Phase A~D + Phase 2 전체 반영
...
## IMPROVEMENT (Phase A~D)
- A-1: 4단계 Sonnet 디자인 조정 (_adjust_design) — CSS 변수 cascade
- A-2: 5단계 HTML 전문 프롬프트 전달
- A-3: shrink/expand 하드코딩 제거 → Sonnet target_ratio 기반
- A-4: rewrite action 구현
- A-5: overflow: visible (area 레벨 텍스트 잘림 방지)
- A-6: object-fit cover → contain (이미지 crop 방지)
- A-7: table-layout: fixed
- A-8: container query 폰트 스케일링
- B-1: details-block 템플릿 신규 (CSS 변수만 사용)
- B-2: 인쇄 시 details 자동 펼침 JS
- B-3: catalog에 details-block 등록
- B-4/B-5: images[]/tables[] 상세 판단 + fallback 3곳 동기화
- B-8: fallback card-grid → topic-header + char_guide 제거
- C-1: CLAUDE.md gradient 원칙 완화
- C-3: border-radius 9개 파일 var(--radius) 통일
- C-4: box-shadow 2레벨 → 1레벨
- D-0: 이미지 경로 입력 UI + API base_path
- D-1: Pillow 의존성 + image_utils.py
- D-2~D-4: 이미지 비율/축소방지 프롬프트 전달
- D-5: HTML에 이미지 base64 삽입
## Phase 2 (다른 Claude 작업)
- P2-A: FAISS 블록 검색 (bge-m3, 46개 블록)
- P2-B: SVG N개 자동 배치 (svg_calculator.py)
- P2-C: Opus 블록 추천 (Kei API 경유)
- P2-D: 5단계 재검토 루프 강화 (MAX_REVIEW_ROUNDS=2)
- P2-E: details-block fallback 연동
## 버그 수정 (BF-8~10)
- BF-8: 컨테이너 예산 기반 블록 배치
- BF-9: grid와 Sonnet 역할 분리
- BF-10: catalog mtime 캐시 자동 갱신
## 블록 라이브러리
- 46개 블록 (6 카테고리), catalog/BLOCK_SLOTS/INDEX 동기화
- 구 블록 제거 (quote-block, card-grid, comparison)
- 13개 _legacy 블록 보존
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-03-25 18:40:20 +09:00
