Stats
Actions
Tags
From flow
Push the task branch, open a Korean pull request, then hand off to flow:code-review for the multi-LLM review pipeline. Use this whenever development for a flow task is done and the branch is ready for review. Even when the user says "just open a PR", run the full flow — multi-LLM review is the point of this skill, not optional dressing. Run in its own session from develop; do not bundle.
How this skill is triggered — by the user, by Claude, or both
Slash command
/flow:deployThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Deploy is the closing skill. It is intentionally separate from develop so reviewer LLMs see a clean diff without develop's intermediate state in context.
Deploy is the closing skill. It is intentionally separate from develop so reviewer LLMs see a clean diff without develop's intermediate state in context.
The review pipeline itself lives in flow:code-review, which deploy invokes after opening the PR. Splitting the two means the review skill is reusable for arbitrary existing PRs (not just ones deploy just created).
tasks.md are checked.If any precondition fails, stop and tell the user. Do not "fix it up" silently.
git push -u origin "$(git branch --show-current)"
If the push fails because of upstream changes, do not force-push. Inform the user and ask whether to rebase.
PR title: short, ≤ 70 chars, conventional-commit-flavored (feat(auth): Google 로그인 지원 추가).
PR body in Korean, using this template:
## 개요
<1-3 줄 요약 — 무엇이 왜 바뀌었는지>
## 변경 사항
- ...
- ...
## 테스트
- [ ] Vitest 유닛/통합 테스트
- [ ] Playwright E2E (해당 시)
- [ ] 수동 확인: <어떤 시나리오를 어떻게 확인했는지>
## 스크린샷 / 영상
(UI 변경이 있는 경우)
## 관련 링크
- 플랜: `.planning/<date>-<task>/plan.md`
- 리뷰 요약: `.planning/<date>-<task>/artifacts/code-review-summary.md` (자동 생성 예정)
Create with HEREDOC for correct formatting:
gh pr create --title "<title>" --body "$(cat <<'EOF'
<body>
EOF
)"
Capture the PR number from the URL gh pr create prints.
Invoke flow:code-review with the PR number from Step 2. Because the current branch matches the PR's head and .planning/<date>-<task>/ exists, code-review will resolve the output directory to .planning/<date>-<task>/artifacts/ automatically — same layout as before.
Tell the user:
PR # opened. Handing off to flow:code-review for the multi-LLM review pipeline.
Then invoke the skill. Code-review handles diff gathering, parallel reviewer dispatch, Korean summary synthesis, inline comment posting, and committing the review artifacts. Deploy's job ends once code-review takes over.
flow:code-review precisely so it can be reused for existing PRs and so the multi-LLM dispatch contract has a single source of truth. Don't paraphrase or re-implement it here.flow:code-review directly — not deploy.../code-review/SKILL.md../../references/multi-llm.md../../references/inline-review-posting.md../../references/frontmatter.md../../references/models.md../../references/doc-style.mdGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub gitgitwi/council-flow --plugin flow