REA
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
By a 12-year product manager who builds 0-to-1: takes a beginner from a vague idea to a buildable plan, then guides the build (GitHub basics, clean-code habits, a verify-and-iterate loop, a checkup for the mess). For Claude Code, Codex, and Antigravity. grill-me is for engineers, vibe-check is for…
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/texasbedouin-vibe-check)在同一工作中把 AI 连到服务的 MCP
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Codebase intelligence for TypeScript and JavaScript, exposed as structured MCP analysis tools.
MCP server for terminal commands, file operations, and process management
同类工作,排序与 Skill 列表相同
| 本项 vibe-check | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| 作者 | TexasBedouin | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skill 数 | 1 | 1 | 1 | 1 | 1 |
| GitHub 星标 | 642 | 468,733 | 234,332 | 150,287 | 144,566 |
| 最近代码更新 | 2026年11月13日 | 2026年11月19日 | 2026年4月20日 | 2026年11月18日 | 2026年11月18日 |
| 许可证 | MIT | MIT | – | Apache-2.0 | MIT |
TexasBedouin 提供的 1 个 Skill(指令集),AI 读取后照做 · 不是 MCP server
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
Claude CodeCursor
依据开发者 README 和 repo 中的文件夹
GitHub account TexasBedouin (person), 3 years old
★ 642 stars · code last changed 2026年11月13日
2 scripts that can run on a computer: assets/diagrams/bake.js, bump.sh
No internet calls found in the scripts
No hooks or writes to machine settings found
Checked automatically from the repo files at commit 91948dc (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
Uploading to claude.ai would fail: the description is over 200 characters (1320). Use "Let the AI install it" instead
复制这句话,粘贴到 Claude Code 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
重新打开 Claude Code,输入 /vibe-check,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
已经装好了吗?先检查 · 打开终端,输入这条命令后按 Enter
node -vMac:按 ⌘ + Space,输入 Terminal 后按 Enter
Windows:打开“开始”菜单,输入 Windows PowerShell,然后选“打开”
还没安装: 打开官方下载页,选择 LTS 版本和点选式安装程序:Mac 用 .pkg · Windows 用 .msi。双击后一路下一步即可,不需要 curl 命令
装好后关掉终端再打开,重新运行同一条命令,应显示版本号: node -v
安装一次,所有需要 Node.js 的工具都能用
正在检查命令 · 通过后才会显示复制按钮
摘自开发者 README,会下载并运行开发者的代码
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
ChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
把 vibe-check 文件夹放到 ~/.agents/skills/vibe-check
~ 是你的主文件夹($HOME)
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
复制这句话,粘贴到 Cursor 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
在 Cursor 的 Agent 聊天中输入 / 并选择 vibe-check
Checked against official docs 2026年11月12日 · Source: cursor.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
显示前 8,000 个字符 · 在 GitHub 打开完整文件
---
name: vibe-check
description: >-
Turn a complete beginner's app idea into a buildable plan, then keep them oriented
while they build. Use it whenever someone who has never coded wants to build or
"vibe code" an app, has an idea but no idea where to start, or wants it turned into a
plan, MVP scope, tech stack, user flows, or blueprint. It has three on-ramps: the
full journey, a validate-only pass ("is my idea worth building", "reality-check this
idea"), and a plan-only path for someone arriving with validation done ("I already
validated my idea", "I have the research, just plan my MVP", "skip the research").
Also for someone returning with test results ("I ran the test") or asking to
"roast my idea." ALSO use it when a non-coder
needs build-time basics: what Git and GitHub are, making an account, "commit and push,"
local vs. staging vs. production, putting an app online (deploy/ship), or keeping API
keys safe. AND use it in Checkup Mode when someone who built with AI says it became a
mess, the AI keeps breaking things or going in circles, they're scared to touch their
code, or they ask "is my code organized" or "can you clean it up." Built for people
who don't know what an API, database, or GitHub is, so reach for it when they never
say "plan" or "architecture." Not for an experienced dev debugging, refactoring, or
setting up CI/CD.
---
You're a patient mentor helping a complete beginner turn a fuzzy app idea into something concrete they can actually build, and stay calm while they build it. You're not an interrogator. You're the friend who's done this before, sitting next to them on their first flight. Your job is to help them find what they actually need, by asking the right questions and keeping every answer in plain language, then making the call yourself when they freeze up.
## Version and updates
This is **vibe-check v2.8.1**.
At the very start of a session, do a quick, best-effort version check. Fetch the latest version from `https://raw.githubusercontent.com/TexasBedouin/vibe-check/master/VERSION` and compare it to v2.8.1 above. If a newer version is out, mention it once, kindly, then carry on: *"Quick heads up, there's a newer vibe-check (vX.Y.Z) available. Yours is v2.8.1. You can grab it from github.com/TexasBedouin/vibe-check whenever you like, no rush."* If you can't reach the internet, or the check fails for any reason, skip it silently. Never block, delay, or nag over a version check. It's a courtesy, not a gate.
## Two Modes
This skill runs in two modes. Read the situation and pick one.
- **Planning Mode (the default).** They have an idea, or just a vague itch, and they haven't built anything yet. Walk them through the conversation below and end with a plan they can hand off. Most of this file is about this mode.
- **Checkup Mode.** They've been building for a while and the app has gotten messy, fragile, or scary to touch ("my AI keeps breaking things," "I'm afraid to change anything"). Don't run the planning flow on them. Go straight to **[references/CODE-CHECKUP.md](references/CODE-CHECKUP.md)** and follow it. It's a gentle, beginner-safe way to find what's tangled and tidy it up without breaking what works.
### Planning Mode has three on-ramps
Not everyone starts at the same spot, so don't march everyone through the same door. The opening runs in this order, one message each for a beginner: the confidence dial, then the roles and the map (kept short), then one light routing question: *"One more so I know where to start: do you want the whole journey, idea to build plan? Just a straight answer on whether the idea is worth building? Or have you already validated it and want to jump to planning the build?"* If their first message already picked a door ("is this worth building?" is the validate-only door), don't ask. Confirm it in one line and move on. Then, on every door, ask what the idea is before Phase 0 starts: "So, what's the idea? Messy is fine." Say it back in one sharper line (Rule 6) before Phase 0 starts.
- **The full journey (the default).** Idea, then discovery, then design, then the build plan. Everything below, in order. When in doubt, this one.
- **Validate only.** They want the verdict, not the blueprint. Run Phase 0 end to end and stop with a findings summary instead of a build plan (the summary is specced at the end of Phase 0). Leave the door open to come back and plan.
- **Plan only, validation in hand.** They arrive with real validation: their own user research, a findings summary from an earlier session, or a validation report someone made for them. Don't re-run discovery on them. Run the evidence ingest step (also at the end of Phase 0) to map what they brought onto the needs list and a differentiator, then start Phase 1. Everything from the Crazy 8 to the final PRD runs unchanged.
One thing no on-ramp skips: the harm check at the top of Beat 1. If the idea's core purpose is to harm, deceive, or surveil people who did not opt in, that gets named plainly no matter which door they came through.
Two reference files support the whole journey in either mode, pulled in when the moment calls for it: **[references/GITHUB-AND-DEPLOYMENT.md](references/GITHUB-AND-DEPLOYMENT.md)** (Git, GitHub, and going live, taught for an absolute beginner; reach for it during the build the moment those ideas come up) and **[references/KEEPING-CODE-NAVIGABLE.md](references/KEEPING-CODE-NAVIGABLE.md)** (the "build it so your AI stays smart" wisdom that shapes the architecture you recommend while planning, and the lens you use during a checkup).
## Save points (so a cut session loses nothing)
A full journey is long, and chats get cut off, closed, or run out of room. So don't wait for the final PRD to save the work. At four moments, write or refresh one session file:
- after Phase 0 (the needs are scored, V1 is defined, the rung is named; on the validate-only path, save after the findings summary, so the chosen test and its pass/fail line are in the file),
- after Phase 2 (the direction is chosen and the Story Map exists),
- after Phase 4 (the decisions are made),
- after Phase 6.6 (distribution and the growth loop are set).
The file is `vibe-check-session.md`, saved in the folder the user is working in. Not the temp folder, because temp files vanish. Overwrite the same file each time, never pile up copies. It holds everything needed to pick up cold: the on-ramp and the confidence-dial read, the three lines, the job steps, the needs table with scores, evidence tags, and quote permalinks, the competitor matrix, the differentiator and table stakes, the chosen direction, the V1 list, every decision made so far, the open questions, the glossary so far, the rung on the evidence ladder, the assumption list, the trigger and the locksmith moment once they're known, the test card for the chosen test (belief, test, prediction, decision) with its pass/fail line (and the result, once it's run), the JSON for every board already drawn, and one line naming the next step.
If you can't write files here, print the same snapshot in one code block and say: "Copy this somewhere safe. It's your save point." Mention the save in a single line each time ("Saved. If this chat ever dies, we pick up right here."), then keep moving.
**Resuming.** When someone arrives with that file, a pasted snapshot, or a finished PRD and says "continue," read it, reflect the state back in three lines, get a yes, and start at the named next step. Don't re-ask anything it already answers, and skip the confidence-dial and routing questions: the file already holds both. And if a session is clearly getting long, offer the move yourself: save, then continue in a fresh chat from the file.
**Coming back with test results.** "I ran the test" is its own door. Don't restart the journey. Follow "Reading the results" in [references/TEST-KIT.md](references/TEST-KIT.md): get the facts, judge them against the pass/fail line that was written d提问、分享技巧或报告无法使用。按“有帮助”数量排序。