About clawfetch
clawfetch is a web intelligence MCP server for AI agents, with tools to fetch pages, render dynamic sites, extract structured data, and research topics across sources. Tool calls use x402 payments in USDC on Base; its public hosted endpoint is retired, so it must be run locally or accessed through the API.
Key features
- 01Fetch web pages as Markdown
- 02Render JavaScript-heavy pages
- 03Extract structured data from many site types
- 04Research topics across sources with citations
- 05Check domain availability and enrich company profiles
- 06Pay with USDC through x402
Use cases
- 01Fetch page content for agent analysis
- 02Research a topic and gather cited sources
- 03Check domain availability and suggestions
- 04Build a company profile from its domain
MCP support
This MCP server provides compatible clients with web fetching, rendering, data extraction, and research tools. It can run locally over stdio or Streamable HTTP, with tool calls paid through x402 using USDC on Base.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y @clawfetch/mcpCLAWFETCH_PRIVATE_KEYSecretEVM private key for a Base wallet holding USDC (pays x402 micropayments gaslessly)Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project README@clawfetch/mcp
MCP (Model Context Protocol) server for ClawFetch — Web Intelligence API for AI Agents.
Gives any MCP-compatible client (Claude Desktop, Cursor, Windsurf, OpenClaw, etc.) instant access to ClawFetch's web intelligence tools: fetch pages as markdown, render JS-heavy SPAs, extract structured data from 17+ site types, research topics multi-source, enrich companies from a domain, and check domain availability.
Payments are handled automatically via the x402 protocol — USDC on Base. No API keys, no billing dashboards.
Access and payment
The public hosted MCP endpoint is retired and returns HTTP 410 Gone. Run the MCP locally using the setup below, or call https://api.clawfetch.ai directly. Paid tool calls use your own USDC-funded wallet on Base; ClawFetch does not subsidize requests.
💻 Self-Hosted Mode
Run your own instance with full control over the payment wallet.
Install
npm install -g @clawfetch/mcp
Configure (stdio — for local MCP clients)
Add to your MCP client config (e.g., Claude Desktop):
{
"mcpServers": {
"clawfetch": {
"command": "clawfetch-mcp",
"env": {
"CLAWFETCH_PRIVATE_KEY": "0xYOUR_ETHEREUM_PRIVATE_KEY"
}
}
}
}
The private key signs x402 micropayments (USDC on Base). Fund the corresponding address with USDC on Base mainnet.
Run as HTTP Server
Expose the MCP server over HTTP for remote clients:
CLAWFETCH_PRIVATE_KEY=0x... CLAWFETCH_TRANSPORT=http CLAWFETCH_PORT=3001 clawfetch-mcp
This starts a server at http://localhost:3001/mcp using the Streamable HTTP transport (MCP spec compliant). By default it binds to 127.0.0.1 and accepts only loopback Host headers, which blocks DNS-rebinding attacks from web pages. To serve remote clients, set CLAWFETCH_HOST=0.0.0.0 and list your public hostnames in CLAWFETCH_ALLOWED_HOSTS.
HTTP Endpoints
| Endpoint | Method | Description |
|---|---|---|
/mcp | POST | Initialize session — returns Mcp-Session-Id header |
/mcp | GET | SSE stream for server-to-client notifications (requires session ID) |
/mcp | DELETE | Close a session |
/health | GET | Server health check (JSON) |
Session Flow
- Initialize:
POST /mcpwith JSON-RPCinitializerequest → getMcp-Session-Idheader - Call tools:
POST /mcpwithMcp-Session-Idheader + JSON-RPCtools/callrequest - Stream:
GET /mcp?sessionId=...for SSE notifications (optional) - Close:
DELETE /mcpwithMcp-Session-Idheader
Tools
| Tool | Description | Cost |
|---|---|---|
fetch_url | Fetch any URL as clean markdown | $0.001 |
render_page | Render JS-heavy pages (SPAs, React apps) | $0.005 |
extract_data | Extract structured data from 17+ site types | $0.008 |
research_topic | Multi-source research with citations | $0.02 |
check_domains | Check domain name availability | $0.003 |
suggest_domains | Generate available domain suggestions | $0.003 |
parse_document | Parse office docs (docx/pptx/xlsx/pdf/…) to markdown | $0.005 |
enrich_company | Company profile from a domain: socials, emails, tech stack, key pages | $0.02 |
list_extractors | List supported structured data extractors | $0.001 |
health_check | Check API health status | Free |
wallet_info | Show payment wallet address | Free |
Supported Extractors
GitHub repos/profiles, npm packages, PyPI packages, Twitter/X profiles, LinkedIn profiles/companies, YouTube videos/channels, Product Hunt, Hacker News, Reddit, Crunchbase, Wikipedia, and more.
Use list_extractors to see all available types with their schema.
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
CLAWFETCH_PRIVATE_KEY | Yes* | — | Ethereum private key (0x-prefixed) for x402 payments |
CLAWFETCH_BASE_URL | No | https://api.clawfetch.ai | API base URL |
CLAWFETCH_TRANSPORT | No | stdio | Transport: stdio or http |
CLAWFETCH_HOST | No | 127.0.0.1 | HTTP server bind address |
CLAWFETCH_ALLOWED_HOSTS | No | localhost,127.0.0.1,[::1] | Comma-separated Host header allowlist for the HTTP transport (DNS-rebinding protection) |
CLAWFETCH_PORT | No | 3001 | HTTP server port |
CLAWFETCH_TIMEOUT_MS | No | 60000 | Request timeout in ms |
CLAWFETCH_DEBUG | No | false | Enable debug logging (1 or true) |
* Required for tool calls. Without it the server still starts and lists its tools, resources and prompts (so registries can inspect it), and every call returns setup instructions. Keep this key in your local configuration; never submit it to the ClawFetch website.
Development
git clone https://github.com/clawfetch/clawfetch-mcp.git
cd clawfetch-mcp
npm install
npm run build
npm test
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.