@aiwerk/mcp-server-clawhub
Enables browsing, searching, inspecting, downloading, and publishing skills from the ClawHub registry through the Model Context Protocol.
README
@aiwerk/mcp-server-clawhub
MCP server for the ClawHub.ai skill catalog.
Browse, search, inspect, download, and (when authenticated) publish skills from the ClawHub registry through the Model Context Protocol.
Two modes
The server adapts to whether a token is configured:
| Mode | Trigger | Available tools |
|---|---|---|
| anonymous | CLAWHUB_TOKEN unset |
10 read-only tools (search, list, get, scan, moderation, file, resolve, download) |
| authenticated | CLAWHUB_TOKEN set |
All 14 tools (adds whoami, publish, delete, undelete) |
The authenticated tools are simply not registered in anonymous mode — tools/list will not advertise them.
Install
Two ways to run this server — pick the one that fits.
Option 1 — Hosted (zero setup)
No local runtime — if you set a token, it's AES-256-GCM encrypted server-side via HashiCorp Vault.
- Sign up at aiwerkmcp.com.
- Install ClawHub from the catalog. Leave
CLAWHUB_TOKENempty for anonymous mode (10 read-only tools), or paste a token to unlock all 14. - Point your MCP client (Claude.ai, Cursor, Hermes, …) at your hosted endpoint:
with your Bearer token.https://bridge.aiwerk.ch/u/<your-user-id>/mcp
Option 2 — Self-hosted (npx)
Run directly — you manage the token:
npx -y @aiwerk/mcp-server-clawhub
Or in your MCP client config:
{
"mcpServers": {
"clawhub": {
"command": "npx",
"args": ["-y", "@aiwerk/mcp-server-clawhub"],
"env": {
"CLAWHUB_TOKEN": "optional - unlocks publish/delete/whoami"
}
}
}
}
Tools
Read-only (anonymous + authenticated)
| Tool | Description |
|---|---|
clawhub_search |
Search skills by query string — primary discovery path |
clawhub_list_skills |
List skills with cursor pagination (see caveat below) |
clawhub_get_skill |
Full details for a skill slug |
clawhub_list_versions |
List all versions of a skill |
clawhub_get_version |
Get a specific version (files + security snapshot) |
clawhub_get_scan |
Security scan result for a version |
clawhub_get_moderation |
Moderation verdict and evidence (response wraps under moderation) |
clawhub_get_file |
Fetch a single raw file from a skill |
clawhub_resolve |
Resolve version by content hash |
clawhub_download |
Download skill zip (base64 encoded) |
Discovery caveat: the live
/skillsendpoint applies a default server-side filter that may return an emptyitems[]for ungated browsing. In practice, preferclawhub_search— it's the reliable entry point for finding skills by keyword. Useclawhub_list_skillswith cursor pagination only when you have a specific filter in mind (e.g.nonSuspiciousOnly=true). ThehighlightedOnlyflag is available onclawhub_searchbut not onclawhub_list_skills.
Wire shapes worth knowing
Some live API responses differ from the published OpenAPI spec. The types in src/types.ts follow
the real wire shape:
clawhub_list_skills→{ items: [...], nextCursor }(notskills)clawhub_list_versions→{ items: [...], nextCursor }(notversions)clawhub_get_moderation→{ moderation: { verdict, ... } }(wrapped)clawhub_whoami→{ user: { handle, ... } }(wrapped)clawhub_get_moderationreturns 404 for skills with no moderation events — this means clean, not missing. For the always-present verdict, read.moderationfromclawhub_get_skill.
Authenticated only (CLAWHUB_TOKEN required)
| Tool | Description |
|---|---|
clawhub_whoami |
Current user (handle, displayName, image) |
clawhub_publish |
Publish a new skill version |
clawhub_delete |
Soft-delete a skill (reversible) |
clawhub_undelete |
Restore a soft-deleted skill |
Environment
| Variable | Required | Purpose |
|---|---|---|
CLAWHUB_TOKEN |
no | Bearer token issued by clawhub.ai. Unlocks auth-only tools and raises rate limits (120/min → 600/min for reads). |
CLAWHUB_BASE_URL |
no | Override the API base URL. Defaults to https://clawhub.ai/api/v1. |
Rate limits
- Anonymous: 120 reads/min per IP, 30 writes/min per IP
- Authenticated: 600 reads/min per key, 120 writes/min per key
Development
npm install
npm run build
npm test
About AIWerk MCP
Part of the AIWerk MCP platform — curated, signed MCP recipes served either as npm packages for self-hosting or through our multi-tenant hosted bridge (bridge.aiwerk.ch).
Other AIWerk MCP servers:
- @aiwerk/mcp-server-cal — Cal.com scheduling
- @aiwerk/mcp-server-imap — IMAP/SMTP email, provider-agnostic
- @aiwerk/mcp-server-wise — Wise (TransferWise) Personal API, read-only
Browse the full catalog (20+ recipes including GitHub, Linear, Notion, Stripe, …) at aiwerkmcp.com.
License
MIT © AIWerk
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。