All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
1.14.2 (2026-04-26)
- commands: translate /pf:help body to English (partial #113) (#118) (6957022)
- manifest: translate marketplace.json plugin description to English (#112) (#115) (f91571e)
1.14.1 (2026-04-26)
1.14.0 (2026-04-26)
1.13.0 (2026-04-25)
- workflow: H1->SpecDD auto-advance — sentinel hook + dispatch script + M3 imperatives (Phase 1) (#103) (17f354a)
- workflow: H2→preview server auto-launch + /pf:preview slash command (Phase 2) (#104) (e874019)
1.12.1 (2026-04-25)
- security: broader eval regex + fake-shim for positive URL loop (#95 / #85 follow-up) (#97) (303c302)
- security: defense in depth — input-path cap + cache read-path size check (#95 / #83 follow-up) (#96) (2136c56)
1.12.0 (2026-04-25)
- ideation,meta: A-4 filled-ratio gate + A-5 H1 modal helper (#59) (#89) (7abe606)
- ideation,meta: A-6 framework lint + C-5 spec-anchor audit (#59 #62) (#88) (07c4182)
1.11.1 (2026-04-25)
- cache: atomic strong+weak alias via hardlink to eliminate concurrent put race (#70) (#81) (798ae02)
- hygiene bundle — verify-plugin schema count + idea_summary cap + macOS dup gitignore (#63 #65 #68) (#83) (93123ec)
- security: SHELL_BYPASSES coverage + regex weakening probes (#64 #69) (#85) (9a2413b)
1.11.0 (2026-04-25)
- v1.7.0: Phase 8 — Q-9 / Q-1 / Q-2 + verifications + Q-4/Q-6/Q-7/Q-8 deferred — closes umbrella #37 (#55) (99a3aa2)
1.10.0 (2026-04-25)
1.9.0 (2026-04-24)
1.8.1 (2026-04-24)
1.8.0 (2026-04-24)
1.7.0 (2026-04-24)
1.6.1 (2026-04-24)
- cache: pre-Socratic weak-key probe + dual-store for one-click replay (A-1, v1.6.1) (#28) (8186455)
- security: reject traversal mockup_path + URL injection in browser opener (S-1/S-2, v1.6.1 hotfix) (#27) (13b83b4)
1.6.0 (2026-04-24)
1.5.4 (2026-04-23)
1.5.3 (2026-04-23)
1.5.2 (2026-04-23)
1.5.1 (2026-04-23)
1.5.0 (2026-04-23)
- ci: add template-build smoke test + content checks (B3 fix) (#11) (5a4d1d4)
- scc: add build_config + template_gap categories + scc-build-config fixer (B4) (#12) (d9a5d00)
1.4.0 (2026-04-22)
- assets: Phase O — standard-profile templates (SQLite + dockerless MVP) (1a86e6c)
- hooks: Phase Q — escalation decision ledger + pre-flight integration (18de1ab)
- profiles: Phase N — default=standard + stack hints + escalation config (a41a97f)
- scripts: Phase P — recommend-profile.sh (categorical signal scoring) (a98e2ff)
- v1.4.0: default=standard + local-first MVP (SQLite + no-Docker) + profile escalation (c9d73b3)
- review: Phase T — Gemini + Codex review feedback (6 issues) (bfc0574)
- review: Phase T-2 — CodeRabbit re-review feedback (10 issues) (1de4c46)
1.3.0 (2026-04-22)
- agents: Phase D — Proposals #1 #4 #10 (profile-aware previews + budgets + context editing) (fc2c2d6)
- agents: Phase F — Proposals #3 #5 #6 #7 #8 #9 (eng teams, SCC auto-extend, QA tools/agents split) (cec62b5)
- commands: Phase I — /pf:status + /pf:budget profile-aware (39b62de)
- hooks: Phase B — Layer-0 Rule 9 idea-drift detector (P0-A) (abc9695)
- monitors: Phase C — P0-B cost-regression sentinel (926309d)
- profiles: Phase A — standard/pro/max profile schema foundation (91e24d9)
- scripts: Phase E — surface-type detection (Proposal #2) (0f4a1d3)
- scripts: Phase H — PreviewDD-level cache (Proposal #11) (1e041fa)
- v1.3.0: profiles (standard/pro/max) + Rule 9 drift + cost regression + 11 panel-validated changes (e44670c)
- cache: use python for mtime instead of stat (portable across macOS/Linux) (9031fc3)
- review: Phase M — Gemini + Codex review feedback (7 issues) (0851b68)
1.2.1 (2026-04-22)
- release: sync plugin.json + marketplace.json to 1.2.0 + use manifest mode (24b9c0b)
1.2.0 (2026-04-22)
- hooks: Rule 8 — run artifact single-writer enforcement (da643d2)
- ci: shell-safe release-please output debug (env pattern, no shell interpolation of JSON blob) (76b1bb9)
- Gate H1 is now preview-selection + design-tweak (was design-only)
- 첫 real run(r-20260422-184337)에서 사용자가 panel top-3 밖의 P19(legal-depo)를 선택. v1.0.0은 panel 추천만 자동 lock하고 agency를 주지 않았음.
commands/design.mdrewritten: AskUserQuestion 4옵션 (Recommended / Alt A / Alt B / Gallery of 26).agents/meta/chief-engineer-pm.md§3 rewritten: full selection procedure.- Alternative 선택 시 mitigations 재생성 의무화 (제품 context 달라짐).
chosen_preview.panel-recommended.json으로 panel 원본 백업.
memory/LESSONS.md0.7: "Panel 추천 ≠ 사용자 의지" (category 1 PreviewDD, 핵심 결함)
runs/r-20260422-184337/chosen_preview.panel-recommended.json(P02 Slack bot)runs/r-20260422-184337/chosen_preview.json(user-overridden to P19 legal-depo)
Plugin scaffold complete. Ready for /plugin install pf@two-weeks-team via
Two-Weeks-Team/PreviewForgeForClaudeCode marketplace.
- ✅ Phase 0 — Repo scaffold + git init + first push (c351a96)
- ✅ Phase 1 — Marketplace + Plugin manifests (9a035ca)
- ✅ Phase 2 — Memory seed + Layer-0 methodology (4d31045)
- ✅ Phase 3 — 3 hooks (AskUserQuestion enforcement · factory-policy · auto-retro) (ae5497f)
- ✅ Phase 4 — Meta Layer: M1 Run Supervisor · M2 Cost Monitor · M3 Chief Engineer PM (eae47e6)
- ✅ Phase 5 — Ideation Dept (29 agents: I_LEAD + I1/I2 + 26 Advocates) (6d97ea6)
- ✅ Phase 6 — 4-Panel decision (45 agents: 4 chairs + 40 members + MD) (2f9638b)
- ✅ Phase 7 — Spec Dept (9 agents: LEAD + AUTHOR + 7 specialist critics) (c7ffe26)
- ✅ Phase 8 — 5 Engineering Teams (25 agents: BE/FE/DB/DO/SDK leads+members) (b9b6b8f)
- ✅ Phase 9 — QA Dept (14) + Self-Correction Squad (5) (d7a40af)
- ✅ Phase 10 — Judges (5) + Auditors (5) + Docs Squad (3) · 143 agents total (89e341b)
- ✅ Phase 11 — 14 slash commands
/pf:*(653824b) - ✅ Phase 12 — Assets · schemas · seed ideas · monitors · settings · bin (8a1ffea)
- ✅ Phase 13 — Plugin install verified + verify-plugin.sh (34/34 checks pass)
/plugin marketplace add Two-Weeks-Team/PreviewForgeForClaudeCode
/plugin install pf@two-weeks-team
/reload-plugins
/pf:bootstrap
/pf:new "your one-line idea"
- Marketplace clone via GitHub works (✔ Successfully added marketplace)
- Plugin install works (✔ Successfully installed plugin: pf@two-weeks-team)
- All 14
/pf:*commands discoverable (tested/pf:help) - All 143 agents loaded
- All 3 hooks compile + execute correctly
- No monitors load errors
- v1.1 — First end-to-end run (idea → freeze) with a seed idea
- v1.2 — Auto-retro critic operational (LESSONS self-population from real runs)
- v1.3 — Built-in Design Studio UI (currently Gate H1 is spec'd; UI to be implemented at first use)
- v2.0 — Generalization beyond Nestia stack (other spec-first TS stacks)