REA
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
google maps for vibecoders.
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/quinnaho-claudemap)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 claudemap | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| 作者 | QuinnAho | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skill 数 | 1 | 1 | 1 | 1 | 1 |
| GitHub 星标 | 295 | 468,733 | 234,332 | 150,287 | 144,566 |
| 最近代码更新 | 2026年4月21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 |
QuinnAho 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 CodeCodex
依据开发者 README 和 repo 中的文件夹
GitHub account QuinnAho (person), 3 years old
★ 295 stars, few users so far · code last changed 2026年4月21日
60 scripts that can run on a computer: commands/create-map.js, commands/open-claudemap.js, commands/refresh.js, commands/setup-claudemap.js and 56 more
Could not check every script (some are too large); read them first
No hooks or writes to machine settings found
Checked automatically from the repo files at commit c0a9b93 (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
Files are byte-for-byte the same as on GitHub at commit c0a9b93 · View on GitHub ↗
60 scripts
Once uploaded it runs on Claude’s systems, not on your computer. No need to unzip
All 65 files · 296 KB
SKILL.md7 KBcommands/create-map.js11 KBcommands/open-claudemap.js4 KBcommands/refresh.js11 KBcommands/setup-claudemap.js9 KBcommands/show.js19 KBcommands/snapshot.js1 KBlib/active-map.js1 KBlib/cache.js2 KBlib/command-harness/enrichment-io.js1 KBlib/command-harness/log.js1 KBlib/command-harness/parse-args.js5 KBlib/command-harness/project-root.js1 KBlib/command-harness/render-slash-template.js2 KBlib/command-harness/run-command.js5 KBlib/command-harness/slash-only-descriptors.js2 KBlib/command-harness/with-mcp.js1 KBlib/contracts/errors.js3 KBlib/contracts/graph-sources.js1 KBlib/contracts/index.js1 KBlib/contracts/paths.js11 KBlib/contracts/paths.test.js11 KBlib/contracts/presentation.js1 KBlib/contracts/schemas/cache.js1 KBlib/contracts/schemas/graph.js1 KBlib/contracts/schemas/index.js3 KBlib/contracts/schemas/install-record.js2 KBlib/contracts/schemas/manifest.js1 KBlib/contracts/schemas/runtime-envelope.js1 KBlib/contracts/schemas/schemas.test.js11 KBlib/contracts/schemas/shared.js2 KBlib/contracts/versions.js4 KBlib/contracts/versions.test.js9 KBlib/differ.js3 KBlib/differ.test.js12 KBlib/enrichment.js1 KBlib/enrichment/enrichment.test.js11 KBlib/enrichment/graph-validation.js2 KBlib/enrichment/health.js2 KBlib/enrichment/icons.js1 KBlib/enrichment/index.js8 KBlib/enrichment/prompts.js2 KBlib/enrichment/source-priority.js2 KBlib/file-walker.js7 KBlib/import-resolution.js4 KBlib/launcher.js2 KBlib/manifest/fingerprint.js3 KBlib/manifest/index.js2 KBlib/manifest/io.js2 KBlib/manifest/migrations.js2 KBlib/manifest/normalize.js3 KBlib/manifest/scope-resolution.js3 KBlib/manifest/scope-resolution.test.js11 KBlib/map-manifest.js2 KBlib/mcp-client.js24 KBlib/runtime-location.js4 KBlib/runtime-location.test.js4 KBlib/runtime-paths.js3 KBlib/scoped-map.js7 KBlib/show-helpers.test.js8 KBpackage.json1 KBprompts/enrichment.txt5 KBprompts/scoped-enrichment.txt5 KBvitest.config.js1 KBLICENSE1 KB正在检查命令 · 通过后才会显示复制按钮
+Create skillUpload a skillpick claudemap-runtime.zip
SettingsCapabilitiesCode execution and file creation
Turn on Code execution and file creation first (newer menus: Create and edit files) · personal plans: Settings → Capabilities · Team/Enterprise: an admin turns it on under Organization settings → Plugins & skills · every plan, including Free
Upload once; it works in claude.ai and Claude Desktop, and in recent Claude Code signed in with a Claude account (not with an API key)
Cowork and cloud sessions use the skills turned on in your claude.ai account, not ~/.claude/skills on your computer (Claude Desktop scheduled tasks run locally and still load it)
Checked against official docs 2026年11月12日 · Source: support.claude.com, support.claude.com, code.claude.com
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 /skill,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 claudemap-runtime 文件夹放到 ~/.agents/skills/claudemap-runtime
~ 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 skill
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
7,596 个字符 · 在 GitHub 打开完整文件
---
name: claudemap-runtime
description: Internal ClaudeMap runtime for turning a repository into a live architecture map and driving that map during walkthroughs. Prefer the public commands in .claude/commands for normal use.
---
ClaudeMap is a repo-to-architecture-map workflow.
High-level model:
- snapshot the repository
- ask `@claudemap-architect` to turn that snapshot into a detailed, human-legible graph
- render the graph in the bundled ClaudeMap UI
- keep the graph and presentation state updated as the user explores the codebase
Public commands:
- `/setup-claudemap`: build or rebuild the map for the current repository
- `/open-claudemap`: reopen the existing UI without rebuilding
- `/create-map`: create or refresh a scoped subsystem map from the current root graph
- `/refresh`: update the graph after code changes
- `/explain`: run a guided walkthrough through the live map
- `/show`: direct the live map for highlights, focus, presentation, and flows
If this skill is invoked directly, default to the setup workflow.
Target repository:
- If the user passed an argument to the invoked skill command, use `$ARGUMENTS` as the project root.
- If no argument was passed, use the current working directory.
Execution rules:
1. Resolve the bundled ClaudeMap workspace from `${CLAUDE_SKILL_DIR}`.
2. Generate a raw repo snapshot by running `${CLAUDE_SKILL_DIR}/skill/commands/snapshot.js` for the target project root.
3. Read `${CLAUDE_SKILL_DIR}/skill/prompts/enrichment.txt`.
4. Use the `@claudemap-architect` subagent explicitly. Give it:
- the raw snapshot JSON
- the graph schema contract from the enrichment prompt
- instructions to return only valid graph JSON
- instructions to optimize for a detailed graph with intuitive human grouping
5. **Wait for the subagent Task call to fully return**, then save the returned JSON to `${CLAUDE_SKILL_DIR}/tmp/claudemap-enrichment.json`. Do not run the setup JS command until after this file has been written with non-empty valid graph JSON. Do not run setup in parallel with the subagent call.
6. Run `${CLAUDE_SKILL_DIR}/skill/commands/setup-claudemap.js` for the target project root with `--enrichment-file ${CLAUDE_SKILL_DIR}/tmp/claudemap-enrichment.json`. The setup command is strict: if the file is missing, empty, or unparseable it will exit non-zero and refuse to render a heuristic graph. Do not retry setup without fixing the enrichment file first.
7. Add `--force-refresh` only when the user explicitly asks for a fresh rebuild.
8. If the subagent fails to return valid JSON after two attempts, tell the user the architect pass failed and stop. Do not silently rerun setup without `--enrichment-file` — that would render a heuristic graph and pollute the cache. The user should rerun `/setup-claudemap` once the architect issue is resolved.
9. Let the bundled launcher start the app unless the user explicitly asks not to.
10. Summarize the analyzed file count, system count, graph source, render transport, and app readiness. Runtime graph outputs land in `${CLAUDE_SKILL_DIR}/app/public/graph/` (served by the bundled Vite app as `/graph/*`).
11. End the graph-generation flow with a short feedback prompt such as: `Does this map look right, or should I refine it?`
12. If the user says the map is good, stop there.
13. If the user asks for refinement, reuse the current target project's `claudemap-cache.json` graph and file snapshot as context when available instead of starting from a blank prompt again.
14. For refinement passes, send the existing graph plus the user's requested changes back through `@claudemap-architect`, save the refined JSON to `${CLAUDE_SKILL_DIR}/tmp/claudemap-enrichment.json`, and run `${CLAUDE_SKILL_DIR}/skill/commands/refresh.js` with `--enrichment-file` so the graph iterates in place.
15. After the refined graph renders, ask the same short feedback prompt again.
`/create-map` runs an architect-first scoped pipeline:
1. Resolve the target scope from the user's request (either a scope JSON payload copied from the UI, or a natural language description of a subsystem the architect should locate in the root graph).
2. Read `${CLAUDE_SKILL_DIR}/skill/prompts/scoped-enrichment.txt`. This is a dedicated scoped prompt — do not reuse the root enrichment prompt.
3. Build a scoped snapshot by filtering the root graph's file list to the files inside the target subsystem. Include any prior scoped graph from the target map's `cachePath` when one exists so the architect can refine rather than rebuild.
4. Call the `@claudemap-architect` subagent with the scoped snapshot, the scoped prompt, and any user instructions. Instruct it to emit a richer internal breakdown (2-6 internal subsystems, tighter edges, function nodes welcome) and to decide between edit-in-place and rebuild based on the intent of the request.
5. Save the returned JSON to `${CLAUDE_SKILL_DIR}/tmp/claudemap-enrichment.json` and run `${CLAUDE_SKILL_DIR}/skill/commands/create-map.js` with `--enrichment-file` (plus `--scope-json` and optional `--instructions`). The command deletes the tmp file after it reads it.
6. If the architect pass fails, the command can still fall back to `buildScopedGraphFromRoot` by running without `--enrichment-file`, but warn the user the scoped map will be a plain filter view and suggest rerunning with architect enrichment for richer grouping.
7. After the scoped map renders, ask the same short feedback prompt (`Does this map look right, or should I refine it?`).
8. If the user asks for refinement, reuse the scoped map's own cache as prior graph context (via `buildScopedSnapshot`'s `priorGraph` option) and rerun the architect. Save the refined JSON to the tmp file and rerun `create-map.js` with the same `--scope-json` payload plus `--enrichment-file`.
9. After the refined scoped graph renders, ask the same short feedback prompt again.
`/refresh` is change-aware for scoped maps:
- The root graph is always diffed against the cached snapshot.
- For each scoped map, if none of its files were in the diff and its cached graph was produced by the architect (`meta.source === 'claude-scoped'`), the scoped map is preserved untouched — no rebuild, no filter fallback.
- If the scoped map's files were touched, it is rebuilt from the updated root graph filter and marked `needsRebuild: true` so the next `/create-map` pass knows to rerun the architect for that scope.
- If a scope can no longer be resolved in the new root graph, it is marked `stale: true` and `needsRebuild: true` and left for the user to confirm.
Important details:
- The bundled runtime lives inside this skill directory, so keep all paths anchored to `${CLAUDE_SKILL_DIR}`.
- `/setup-claudemap` should treat the `@claudemap-architect` path as the primary path, not an optional extra.
- The packaged project includes a `claudemap-architect` subagent in `.claude/agents/` for system identification, graph refinement, and human-first graph restructuring.
- If a cached Claude-authored graph already exists, do not replace it with a heuristic regeneration unless the user explicitly asks for `--force-refresh`.
- If the user only wants to reopen the existing map UI, use `/open-claudemap` instead of rerunning setup.
- Follow-up refreshes should use the `/refresh` command shipped in `.claude/commands/refresh.md`.
- Graph refinements should prefer `${CLAUDE_SKILL_DIR}/skill/commands/refresh.js --enrichment-file ...` over rerunning setup so the current graph context is reused.
- `/show` should be treated as a presentation-direction command, not just a low-level transport wrapper.
提问、分享技巧或报告无法使用。按“有帮助”数量排序。