Clawfetch MCP

What Clawfetch MCP does

The Clawfetch MCP MCP server connects MCP-compatible clients to the ClawFetch Web Intelligence API. It gives agents operations for retrieving web pages as markdown, rendering pages that depend on JavaScript, extracting structured records, researching topics across multiple sources, and parsing office documents into markdown.

Its supported extraction targets include GitHub, npm, PyPI, Twitter/X, LinkedIn, YouTube, Product Hunt, Hacker News, Reddit, Crunchbase, Wikipedia, and other site types. The complete extractor list and each extractor’s schema can be requested with list_extractors.

This server belongs in the Search & Data Extraction category. It is suited to agents that need normalized web information or cited research rather than only raw HTTP responses.

How it works

The Clawfetch MCP MCP server can communicate with a local MCP client through stdio, or run as an HTTP service using MCP Streamable HTTP. In HTTP mode, clients initialize a session at /mcp, receive an Mcp-Session-Id, and send later tool calls with that session identifier. A server-sent events stream is available for notifications, and sessions can be closed with DELETE /mcp. The /health endpoint returns server health information.

Paid calls use the x402 payment protocol with USDC on Base. The configured wallet signs the payments, so the corresponding address must be funded on Base mainnet. Listed prices range from $0.001 for fetch_url to $0.02 for research_topic; health_check and wallet_info are free.

Setup and configuration

Install the package globally with npm:

npm install -g @clawfetch/mcp

For a local client, configure the clawfetch-mcp command and set CLAWFETCH_PRIVATE_KEY to the 0x-prefixed Ethereum private key used for x402 payments. For HTTP mode, set CLAWFETCH_TRANSPORT=http; CLAWFETCH_PORT defaults to 3001, and CLAWFETCH_HOST defaults to 0.0.0.0.

Optional settings include CLAWFETCH_BASE_URL, CLAWFETCH_TIMEOUT_MS, and CLAWFETCH_DEBUG. The private key should remain in local configuration and should not be submitted to the ClawFetch website.

Tools and capabilities

The Clawfetch MCP MCP server exposes these tools:

• fetch_url retrieves a URL as clean markdown.
• render_page handles JavaScript-heavy SPAs and React applications.
• extract_data returns structured information for supported site types.
• research_topic performs multi-source research with citations.
• check_domains checks domain availability.
• suggest_domains generates available domain suggestions.
• parse_document converts DOCX, PPTX, XLSX, PDF, and other supported documents to markdown.
• list_extractors lists available structured-data extractors.
• health_check checks API health, while wallet_info displays the payment wallet address.

Limitations and notes

The hosted public MCP endpoint is retired and returns HTTP 410 Gone. Use a self-hosted local instance or call https://api.clawfetch.ai directly. ClawFetch does not subsidize paid requests, so tool usage requires USDC in the wallet on Base. The README documents Claude Desktop, Cursor, and Windsurf as MCP-compatible client examples; OpenClaw is also mentioned but is not included in the supported client slug list for this directory.

Getting started with this Clawfetch MCP MCP server

Always refer to the official documentation for the most accurate and up-to-date information.