菜单
Use when a plan folder created by sf-flow-plan must be executed in a single worktree (flow/<slug>) with TDD per story and a non-optional audit gate before commit, then finalized so the branch is preserved for a PR.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when a plan folder created by sf-pair-plan must be executed milestone-by-milestone in a worktree, with a per-milestone TDD→review→commit→tracker loop, then finalized so the branch is preserved for a PR.
Use when a user asks to create a structured implementation plan with milestones, stories, and reviewer approval using pi-subagents.
Execute a single task end-to-end with plan review, implementation review, verification, and one persistent task-plan artifact.
Use when a diff or codebase must receive a CodeRabbit-style audit — 7 finder angles, a dual-blind AND-gate, and fix-apply — returning P0–P3 findings and an APPROVED/REVISE verdict.
Use when a defined flow (.pi/workflows/<name>.yaml) must be run end-to-end with no human gates. Input may be a prompt, a markdown file, a PRD, or a Jira story.
Use when a user asks to create or validate a reusable flow from a declarative agents/phases/loops definition. Interviews the user, writes .pi/workflows/<name>.yaml, and registers /<name>.
| name | sf-flow-implement |
| description | Use when a plan folder created by sf-flow-plan must be executed in a single worktree (flow/<slug>) with TDD per story and a non-optional audit gate before commit, then finalized so the branch is preserved for a PR. |
Reviewer agent at ~/.pi/agent/agents/reviewer.md. Reviewer model resolved by the tool. ONE worktree created at flow/<slug> (git-only; non-git targets skip worktree).
Read ai_plan/<slug>/continuation-runbook.md, story-tracker.md, milestone-plan.md.
Reviewer at ~/.pi/agent/agents/reviewer.md (global, write-once, no model in file). Pass model at dispatch.
(Already created by the tool — cd into it.)
For each story: mark in-dev in story-tracker.md → write failing test → red → implement → green → refactor → lint/typecheck/test → commit locally (do NOT push). Review loop per milestone: write diff+verification to /tmp/flow-m<M>.diff, dispatch reviewer (Agent({ subagent_type: "reviewer", model: "<reviewer_model>" })), parse verdict, fix P0/P1/P2, re-dispatch until APPROVED.
Run sf-flow-audit on the accumulated diff. On REVISE (any P0/P1/P2): loop back to the failing STORY (not the whole plan), re-implement, re-audit. Bounded by audit.max_rounds (default 5). P3: fix inline when cheap, else note. (Mirrors bigpowers build-epic step 6 → step 4.)
cd back to main checkout, call sf_flow_finalize (removes worktree dir, preserves flow/<slug> branch). Send Telegram summary.
Update story-tracker.md before/after each story. Commit hash in Notes.
