feat(claude): ddalggak ULW 하위 명령 흡수

[JeremyDev87]

요약

• Claude Code에서 ULW를 독립 /omo-ulw skill이 아니라 /ddalggak ulw-plan, /ddalggak ulw-loop, /ddalggak ulw-research 하위 명령으로 사용하도록 흡수했습니다.
• ddalggak setup은 Claude Code에 skills/ddalggak만 설치하며, 별도 skills/omo-ulw payload는 설치/패키징하지 않습니다.
• ULW/GJC 하위 명령 계약, projection, package manifest, verifier 표면을 ddalggak runtime asset 흐름에 맞춰 갱신했습니다.

사용 표면

/ddalggak ulw-plan "..."
/ddalggak ulw-loop "..."
/ddalggak ulw-research "..."

CLI 출력 검증 표면:

node bin/ddalggak.js ulw-plan --print "테스트 계획"
node bin/ddalggak.js ulw-loop --print "테스트 구현"
node bin/ddalggak.js ulw-research --print "테스트 조사"

각 명령은 각각 /ddalggak ulw-plan ..., /ddalggak ulw-loop ..., /ddalggak ulw-research ... slash command를 출력합니다.

검증

• npm run verify → passed
• node bin/ddalggak.js ulw-plan --print "테스트 계획" → /ddalggak ulw-plan "테스트 계획"
• node bin/ddalggak.js ulw-loop --print "테스트 구현" → /ddalggak ulw-loop "테스트 구현"
• node bin/ddalggak.js ulw-research --print "테스트 조사" → /ddalggak ulw-research "테스트 조사"
• npm pack --dry-run --json --ignore-scripts 경로 검증은 npm run verify의 package verifier에서 통과
• .omo/ evidence/runtime 작업 산출물은 패키지 payload에 포함하지 않음

Evidence Contract

• Required evidence:
  • 하위 명령 print surface가 /ddalggak ulw-* 형태로 출력됨
  • setup/package verifier가 standalone /omo-ulw payload 없이 통과함
  • generated projection/token budget/package manifest drift가 없음
• Evidence not applicable:
  • UI/browser evidence: CLI/Claude Code slash command packaging 변경이라 해당 없음
• Blocking evidence gaps:
  • 없음

수동 머지 경계

이 PR은 current-head checks, review evidence, mergeability를 별도로 확인한 뒤에만 ready/manual merge 판단을 해야 합니다. 자동 merge, release, publish, tag 작업은 이 PR 범위가 아닙니다.

[JeremyDev87]

REVIEW_DONE PR#419 current-head review

Verdict

APPROVE - blocker count 0. This is a top-level self-review verdict, not a formal GitHub approval.

Current head

• head SHA: bd116a220b21340c583aa92f54339e149b501a6d
• base: master
• mergeStateStatus: CLEAN
• review threads/comments: none found

Quality Lens Router Output

• Applicable gate families:
  • simplicity-deletability: setup installer/package surface adds code shape.
  • evidence-contract: readiness/review claim.
  • security-posture: package verification includes repository posture gates; no workflow/release diff in this PR.
• Skipped gates:
  • frontend/react/vercel/mobile gates: not applicable; CLI/package-only change.
  • regression-library: no repeated Medium/High failure pattern found.
• Repo/product conventions that outrank generic rules:
  • Keep status --local scoped to ddalggak parity; do not expand to multi-skill parity in this PR.
  • Keep .omo/ evidence out of package/runtime payload and out of this PR commit.

Evidence Contract

• Required evidence:
  • npm test -> 64/64 passed
  • npm run verify:projections -> passed
  • npm run verify:codex-skill -> passed
  • npm run test:verify-package -> 5/5 passed
  • node scripts/verify-package.mjs -> [verify-package] passed
  • node bin/ddalggak.js setup --dry-run -> /ddalggak and /omo-ulw install actions observed
  • temporary CLAUDE_HOME install -> both skill dirs created with SKILL.md, .installed-version, .installed-manifest.json, then cleaned up
  • npm pack --dry-run --json --ignore-scripts -> claude-skills/omo-ulw/SKILL.md included and .omo/ excluded
  • git diff --check -> passed
  • GitHub checks -> all terminal success on current head
• Blocking evidence gaps:
  • none

Wiki Context Manifest

• Queries attempted: ddalggak setup Claude Code skill omo-ulw ULW
• Wiki sources read: domains/harness-engineering/ddalggak-operating-principles.md
• Relevant wiki facts: completion must be verified through GitHub PR/check/review state, not local tests alone.
• Raw/imported/alias sources seen but not used as authority: harness-engineering/ddalggak-operating-principles.md redirected to canonical path.
• Authority conflicts or evidence gaps: qmd search hung and was aborted; canonical page plus live PR evidence were sufficient.

Manual merge boundary

Ready for manual merge review. No merge or auto-merge was performed.

[JeremyDev87]

REVIEW_BLOCKED PR#419 current-head review

Verdict

BLOCKED - 코드 diff 자체의 신규 Critical/High finding은 없지만, 현재 GitHub PR 상태가 ready/approve 조건을 만족하지 않습니다. 이 코멘트는 ddalggak evidence review comment이며 GitHub formal approval이 아닙니다.

Current head

• PR: feat(claude): ddalggak ULW 하위 명령 흡수 #419
• head SHA: 70f972287f941236976efe7ef088260ae44adba3
• base: master
• branch: feature/omo-ulw-claude-skill
• draft: false
• label: feat

Scope reviewed

• ddalggak/.codex/skills/ddalggak subcommand surface에 ulw-loop, ulw-plan, ulw-research, gjc-plan, gjc-execute, gjc-team 추가
• command contracts, token budgets, projection/package manifests, setup/dispatch tests, README/package verification surface 갱신
• standalone omo-ulw 설치가 아니라 ddalggak 내부 subcommand routing으로 흡수하는 방향 확인

Validation run on current head

• git fetch --prune -> passed
• git rev-list --left-right --count @{upstream}...HEAD -> 0 0
• git diff --check origin/master...HEAD -> passed
• npm run verify -> passed
  • smoke: 67/67
  • verify-package: passed
  • projection verifier: passed
  • codex skill verifier: passed
  • readiness eval: 29/29
  • security posture/workflow gates: passed
  • npm pack dry-run artifact inspection: passed

Blocking GitHub state

• mergeStateStatus: DIRTY
• gh pr checks 419: no checks reported on feature/omo-ulw-claude-skill
• reviewDecision: empty

Review findings

• Critical: 0
• High: 0
• Medium: 0
• Low: 0

Required next action

Resolve the PR's dirty merge state against the current origin/master, rerun npm run verify, push the resulting head, then rerun ddalggak review on the new live PR head. I did not perform merge, reset, rebase, auto-merge, release, or publish actions in this review pass.