菜单
Structured sprint retrospective — what worked, what didn't, action items. Reads harness results + telemetry + git log to generate data-driven retro. Use when: end of sprint, after shipping, or before starting a new feature cycle. Wraps: node bin/vibe.js retro.
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
The primary orchestrator for the 10-phase Vibe-Stack pipeline. Use for: starting projects, planning features, and managing the end-to-end lifecycle. Phases: think, plan, break, build, harness, review, ship, retro, learn, evolve. Maintains state in .vibe/state.json and follows the 'Iron Law' of handoffs.
Auto-evolve rules and retire underperformers — reads telemetry + learnings to propose rule promotions, demotions, and new patterns. Use when: end of sprint, after retro, or when quality scores drift. Wraps: node bin/vibe.js evolve + .vibe/lifecycle/auto-maintain.js.
Production readiness gate — runs 15 automated checks (tests, lint, security scan, skill integrity, catalog validity, state machine). Use when: before shipping, after adding skills, or any time CI should pass. Wraps: node bin/vibe.js harness.
Design audit + anti-slop enforcement + WCAG check + taste-skill dial configuration + impeccable design workflow (init/critique/polish + 18 refinement commands). Use when: reviewing UI code, generating design systems, checking color contrast, applying brand identity, or any time design quality is at stake. Auto-triggers on any UI/CSS/component work. Injects 41 anti-slop rules + 5 impeccable-specific detectors + 3 taste-skill dials + WCAG AA contrast enforcement. Wraps: anti-slop, color-gen, design-system, typography-rules, theme-factory, impeccable-audit skills.
Terse-output mode. Strips filler/hedging from responses to cut output tokens ~50-65% while preserving technical accuracy. Use when: user asks for short/direct/terse answers, wants compressed commit messages or PR comments, or wants token-usage stats. Wraps: caveman-mode skill.
One-click deployment with pre-flight checklist, platform selection, smoke test, and rollback plan. Use when: ready to ship, deploying to staging or production, or setting up CI/CD for the first time. Enforces 14-point done-verifier checklist before any deploy. Wraps: one-click-vercel, one-click-netlify, git-free-deploy, done-verifier skills. Supports: Vercel, Netlify, Railway, Fly.io, AWS, Docker.
| name | vibe-retro |
| description | Structured sprint retrospective — what worked, what didn't, action items. Reads harness results + telemetry + git log to generate data-driven retro. Use when: end of sprint, after shipping, or before starting a new feature cycle. Wraps: node bin/vibe.js retro. |
| argument-hint | [--since <date>] [--save] |
| version | 1.0.0 |
| allowed-tools | ["Bash(node bin/vibe.js retro)","Bash(git log --oneline*)","Bash(cat .vibe/telemetry/*)","Bash(cat .vibe/handoff.md)","Read","Write"] |
--since {date} → limit git log to commits after date (e.g. --since 2026-06-01)--save → write retro findings to .vibe/context-memory.jsonnode bin/vibe.js retro
Reads .vibe/telemetry/harness-results.json + session telemetry.
Run /vibe-learnings --since {date} inline to add pattern/anti-pattern analysis
on top of the CLI retro output.
VIBE-RETRO — {sprint or date range}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PERIOD: {date range} | COMMITS: {N} | HARNESS: {N}/15
WHAT WORKED:
✅ {pattern that paid off}
✅ {tool or approach that saved time}
WHAT DIDN'T:
❌ {approach that caused rework}
❌ {assumption that was wrong}
ACTION ITEMS:
1. {specific change for next sprint}
2. {rule to add or retire}
METRICS:
Fix ratio: {N} fixes / {N} features
Harness pass rate: {N}%
Reverts: {N}
Append findings to .vibe/context-memory.json as type: retro entries.
Available in future sessions via /vibe-status decisions.
Suggest running /vibe-evolve to apply the action items as rule changes.