vibe-check
Skill1 starsChanged 3 months agoMITView on GitHub
【Vibe】意图校准 — 重新对齐「最初想要 vs 现在做出」→ docs/specs/intent-log.md。当用户说「感觉不对/方向偏了」或原型迭代 3-5 轮后使用。
9 other projects use a version of this.
Adopted by
Most recently taken up by
- barkinsarikartal/vibe-checktheir own version, first seen Oct 25, 2026
- raduceuca/vibe-checktheir own version, first seen Oct 20, 2026
- tslateman/skillstheir own version, first seen Oct 11, 2026
- benavlabs/vibe-checktheir own version, first seen Sep 18, 2026
- TexasBedouin/vibe-checktheir own version, first seen Jul 17, 2026
- we-are-singular/vibe-checktheir own version, first seen Jul 17, 2026
Other versions9 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 9
- TexasBedouin/vibe-check★ 581same name
- benavlabs/vibe-check★ 110same name
- allierays/agentic-loop★ 38same name
- barkinsarikartal/vibe-check★ 12same name
- cacheforge-ai/cacheforge-skills★ 10same name
- we-are-singular/vibe-check★ 2same name
- tslateman/skills★ 1same name
- Jinbae10/skills★ 0same name
- raduceuca/vibe-check★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository23 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- studio-arch-designSkill
- studio-discoverySkill
- studio-gate-checkSkill
- studio-project-scanSkill
- studio-reviewSkill
- studio-setup-stackSkill
- studio-sprint-kickoffSkill
- studio-summarize-archSkill
- usds-cost-reportSkill
- usds-debt-logSkill
- usds-micro-adrSkill
- usds-mode-switchSkill
- usds-onboardSkill
- usds-startSkill
- usds-updateSkill
- vibe-branchSkill
- vibe-explain-backSkill
- vibe-graduateSkill
- vibe-prototypeSkill
- vibe-startSkill
- vibe-taste-reviewSkill
The file
Files
技能:Vibe 校准 (/vibe-check)
执行代理: explainer (解说员) 主导 + prompt-refiner (辅助)
概述 (Summary)
Vibe coding 最大的隐性风险是意图漂移:用户不断微调需求,AI 不断跟随, 几轮后偏离最初意图但双方都未察觉。此技能是周期性对齐机制。
触发时机 (Triggers)
- 每完成 3-5 个迭代后(Explorer 主动触发)
- 用户表达"感觉不对"、"不是我想要的"、"我们跑偏了吗"
- 原型代码超过 500 行时(自动提示)
- 用户主动
/vibe-check
工作流 (Workflow)
- 加载上下文:
docs/specs/lite-spec.md(最初意图)docs/specs/intent-log.md(历史校准记录)sandbox/<name>/README.md(当前状态)
- 生成三段式汇报 (Explainer 主导,用非技术语言):
- "你最初说要 X" — 引用 lite-spec 原句
- "我们现在做出了 Y" — 描述当前 demo 实际能做什么
- "两者之间的偏离是 Z" — 明确指出漂移方向
- 展示证据:
- 当前 demo 的运行截图(若前端)
- 或核心功能的输入输出示例
- 封闭式三问 (用
ask_user):- Q1: 整体方向对吗?[对 / 需要 pivot / 直接 kill]
- Q2: 视觉/体验氛围对吗?[对 / 微调 / 重来]
- Q3: 下一步优先级?[功能A / 功能B / 打磨现有]
- 记录到 intent-log: 追加一条 Round 记录
- 产出行动清单: 基于用户回答,给出下一轮迭代的 2-3 条具体动作
成功门控 (Success Gate)
docs/specs/intent-log.md新增了一条 Round 记录- 用户对三个问题都给出了明确回答
- 产出了下一轮的具体行动清单(≤ 3 条)
- 若用户选择
kill或pivot,触发相应流程
输出约束 (Output Budget)
[强制] 完成后返回主上下文只输出:
Vibe 校准完成 (Round [N])。
- 方向判断: [continue / pivot / kill]
- 关键漂移: [一句话]
- 下一步行动:
1. ...
2. ...
3. ...
intent-log 路径: docs/specs/intent-log.md不得在返回摘要中重复三段式汇报正文。
特殊情况处理
- 用户选 kill: 归档
sandbox/<name>/到sandbox/archive/,追加"死因分析"到 intent-log - 用户选 pivot: 建议重跑
/vibe-start并保留 sandbox 作参考 - 连续 3 次 vibe-check 都是 continue 且高一致性: 提示可考虑
/vibe-graduate
协作路径
- continue → 继续迭代
/vibe-prototype或直接编辑 - pivot →
/vibe-start重新对齐 - kill → 归档流程
- 稳定 →
/vibe-graduate晋升
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

