rock-mcp
Documentation crawler MCP server that crawls and indexes documentation sites so that any MCP-compatible AI can search, read, and expand on the content.
README
rock-mcp
Documentation crawler MCP server. Share a link → site gets crawled and indexed → any MCP-compatible AI can search, read, and expand on it.
No API keys. No cloud. Runs locally, stores everything in SQLite.
How it works
- Share a docs URL (
add_docs_url) - rock-mcp crawls and indexes it in the background
- Ask your AI about it — it searches, reads pages, and expands topics on demand
Tools
| Tool | Description |
|---|---|
search_web |
Search the web via DuckDuckGo — find URLs when you don't have one yet |
add_docs_url |
Crawl and index a URL (async, call immediately then search) |
search_docs |
Full-text search across all indexed docs |
get_page |
Get a specific page, chunked. Jump to a section by name. |
expand_topic |
Find all pages covering a topic |
list_sources |
List indexed sources with crawl status and staleness |
recrawl_source |
Re-crawl a source to pick up updates |
get_crawl_status |
Check crawl progress for a URL |
delete_source |
Remove a source and all its pages |
Installation
Prerequisites
- Node.js 18+
- Git
git clone https://github.com/sethwebster/rock-mcp
cd rock-mcp
npm install && npm run build
Note the full path to dist/server.js — you'll need it below.
Claude Code (CLI)
claude mcp add --scope user rock-mcp node /path/to/rock-mcp/dist/server.js
Restart Claude Code after running this. Verify with:
claude mcp list
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"rock-mcp": {
"command": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
}
}
Restart Claude Desktop.
Cursor
Add to .cursor/mcp.json in your project root, or ~/.cursor/mcp.json globally:
{
"mcpServers": {
"rock-mcp": {
"command": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
}
}
Restart Cursor.
Windsurf
Add to ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"rock-mcp": {
"command": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
}
}
Restart Windsurf.
Codex (OpenAI CLI)
Add to ~/.codex/config.yaml (or your project's codex.yaml):
mcp_servers:
- name: rock-mcp
command: node
args:
- /path/to/rock-mcp/dist/server.js
Cline (VS Code extension)
Open VS Code settings, search for Cline MCP, and add:
{
"rock-mcp": {
"command": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
}
Or edit settings.json directly under the cline.mcpServers key.
Continue.dev
Add to ~/.continue/config.json under mcpServers:
{
"mcpServers": [
{
"name": "rock-mcp",
"command": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
]
}
Antigravity
In the agent panel, click ... → Manage MCP Servers → View raw config to open mcp_config.json. Add your server:
{
"mcpServers": {
"rock-mcp": {
"command": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
}
}
Save the file and restart Antigravity (or reload the window).
Firebase Studio (Project IDX)
Add to .idx/mcp.json in your workspace root (create it if it doesn't exist):
{
"mcpServers": {
"rock-mcp": {
"command": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
}
}
For Gemini CLI within Firebase Studio, add to .gemini/settings.json instead.
Zed
Add to ~/.config/zed/settings.json under context_servers:
{
"context_servers": {
"rock-mcp": {
"command": {
"path": "node",
"args": ["/path/to/rock-mcp/dist/server.js"]
}
}
}
}
Any MCP-compatible client
rock-mcp is a standard MCP stdio server. The binary is node /path/to/rock-mcp/dist/server.js. If your client supports MCP, point it there.
Storage
All data lives in ~/.rock-mcp/docs.db (SQLite, WAL mode). Delete it to reset everything.
Crawl defaults
- Depth: 2 hops from the root URL
- Max pages: 50 per source
- Concurrency: 5 pages at a time, 200ms between batches
- Only indexes
text/html— skips PDFs, plain text, JSON, etc. - Redirect deduplication: follows redirects and stores under the canonical URL
License
MIT
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。