← 返回 Skills 市场
AI README Manager
作者
Draco.Cheng
· GitHub ↗
· v1.3.4
· MIT-0
178
总下载
1
收藏
0
当前安装
9
版本数
在 OpenClaw 中安装
/install ai-readme-manager
功能描述
Manages AI_README.md files so AI agents remember your project conventions across every session
使用说明 (SKILL.md)
\r \r
AI README Manager\r
\r
Manages AI_README.md files that document your project's conventions for AI assistants. Gives your agent persistent memory of project rules across every session and every iteration.\r
\r
Setup\r
\r Before this skill works, you must register the MCP server with OpenClaw:\r \r Option 1: CLI (Recommended)\r \r On macOS/Linux:\r
openclaw mcp set ai-readme-manager '{"command":"npx","args":["[email protected]"]}'\r
```\r
\r
On Windows (cmd):\r
```cmd\r
openclaw mcp set ai-readme-manager "{\"command\":\"npx\",\"args\":[\"[email protected]\"]}"\r
```\r
\r
**Option 2: Edit config file**\r
\r
Add to `~/.openclaw/openclaw.json`:\r
```json\r
{\r
"mcp": {\r
"servers": {\r
"ai-readme-manager": {\r
"command": "npx",\r
"args": ["[email protected]"]\r
}\r
}\r
}\r
}\r
```\r
\r
Then restart OpenClaw and verify with `openclaw mcp list`.\r
\r
## When to use\r
\r
Call `get_context_for_file` before any code-related task — editing, creating, or reviewing files in a project.\r
\r
## Tools available\r
\r
- `get_context_for_file` — get relevant project conventions for a specific file path\r
- `init_ai_readme` — initialize and populate empty AI_README files in a project\r
- `update_ai_readme` — record new conventions or architectural decisions\r
- `validate_ai_readmes` — check quality and token efficiency of all AI_README files\r
- `compress_ai_readme` — remove filler language from an AI_README file to reduce token footprint\r
- `discover_ai_readmes` — scan and list all AI_README files in a project\r
\r
## Instructions\r
\r
### MCP: ai-readme-manager\r
\r
Before any code-related task, ALWAYS call `get_context_for_file(projectRoot, path)` first.\r
If `get_context_for_file` reports empty or missing AI_README files, call `init_ai_readme(projectRoot)` to initialize them.\r
If the user's request or your plan conflicts with AI_README conventions (including during planning), STOP and call `update_ai_readme` to resolve the conflict before proceeding.\r
When establishing new conventions or making architectural decisions, call `update_ai_readme` to record them.\r
If AI_README is missing a convention that is already used in 2+ files, call `update_ai_readme` to record it.\r
安全使用建议
This skill appears to do what it claims (manage AI_README.md), but it relies on running a remote npm package via npx and will write to your OpenClaw config and project files. Before installing: (1) review the ai-readme-mcp package source (on its GitHub/npm pages) and check the 1.3.4 release for malicious/unsafe code; (2) prefer installing the package locally or from a pinned, vetted package rather than letting npx fetch it automatically; (3) back up your project and ~/.openclaw/openclaw.json before registering the MCP server; (4) consider running the package in a sandbox or CI runner to inspect its behavior first; (5) limit autonomous invocation if you are uncomfortable with remote execution. If you need more assurance, ask the skill author for a signed release tarball or a reproducible, auditable install path.
功能分析
Type: OpenClaw Skill
Name: ai-readme-manager
Version: 1.3.4
The skill is designed to manage project-specific context for AI agents using 'AI_README.md' files via an MCP server (ai-readme-mcp). The tools and instructions in SKILL.md are consistent with the stated purpose of maintaining architectural conventions and project memory, and there is no evidence of malicious intent, data exfiltration, or unauthorized execution.
能力评估
Purpose & Capability
Name, description, and declared tools (MCP endpoints like get_context_for_file, init_ai_readme, update_ai_readme, etc.) match the stated purpose of managing AI_README.md project conventions.
Instruction Scope
SKILL.md tells the agent to always call get_context_for_file before code tasks and to initialize/update AI_README files when missing or conflicting. This legitimately requires reading and writing files inside the project, and scanning project files to infer conventions. It does not instruct the agent to access unrelated system paths or external endpoints, but it gives the agent authority to create/modify project files and to alter the user's ~/.openclaw/openclaw.json to register the MCP server.
Install Mechanism
The recommended setup registers an MCP server that runs npx [email protected]. Using npx executes code downloaded from the npm registry at runtime, which can run arbitrary code on the host. While downloading an npm package is expected for a Node-based tool, it is a higher-risk install mechanism compared with an instruction-only skill or a vetted local binary. The skill metadata also lists the node package 'ai-readme-mcp', consistent with this behavior.
Credentials
The skill requests no environment variables, credentials, or config paths beyond instructing the user to add an MCP server entry to ~/.openclaw/openclaw.json. There are no unrelated secrets requested.
Persistence & Privilege
always:false (not force-included). The skill requires adding an MCP server entry to the user's OpenClaw config (~/.openclaw/openclaw.json), which persists that integration and allows the agent to invoke the ai-readme-mcp binary via MCP. Combined with the install mechanism (npx), this gives the skill capability to execute remote code whenever the MCP is invoked — consider this when granting persistent registration.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install ai-readme-manager - 安装完成后,直接呼叫该 Skill 的名称或使用
/ai-readme-manager触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.3.4
- Updated MCP registration instructions to reference [email protected].
- Install/setup steps now use the latest package version in all examples.
- Added instruction: If AI_README is missing a convention used in 2+ files, use update_ai_readme to record it.
v1.3.3
- Bumped internal references and installation instructions to use [email protected].
- No functional changes; documentation update only.
v1.3.1
- Added the compress_ai_readme tool to remove filler language and reduce token footprint in AI_README files.
- Bumped version to 1.3.1.
v1.3.0
- Updated MCP server version references from 1.2.10 to 1.3.0 in all setup instructions.
- Minor formatting improvement to the "Instructions" section, adding a heading for clarity.
v1.2.10
- Updated instructions and example commands to reference version 1.2.10 instead of 1.2.9.
- Clarified usage guidance for handling AI_README convention conflicts and user requests.
- No file or functional changes beyond documentation and version update.
v1.2.9
- Bumped MCP server setup instructions to use [email protected].
- Updated version to 1.2.9 in documentation and config examples.
- Added an "MCP: ai-readme-manager" heading above usage instructions for clarity.
v1.2.8
- Updated MCP server installation instructions to reference [email protected] for improved version consistency.
- Bumped skill version to 1.2.8.
v1.2.7
- Added detailed setup instructions for registering the MCP server with OpenClaw, including CLI and manual config file methods.
- No changes to code or functionality; version remains at 1.2.6 despite SKILL.md update.
v1.2.6
Add OpenClaw integration support
- Improved SKILL.md with clearer usage instructions and detailed command descriptions.
- Emphasized when and how to use each tool for maintaining AI_README files.
- Provided guidance for handling missing, empty, or conflicting AI_README situations.
- Updated metadata with installation steps and homepage information.
元数据
常见问题
AI README Manager 是什么?
Manages AI_README.md files so AI agents remember your project conventions across every session. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 178 次。
如何安装 AI README Manager?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ai-readme-manager」即可一键安装,无需额外配置。
AI README Manager 是免费的吗?
是的,AI README Manager 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
AI README Manager 支持哪些平台?
AI README Manager 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 AI README Manager?
由 Draco.Cheng(@draco-cheng)开发并维护,当前版本 v1.3.4。
推荐 Skills