Neura Market
ryoppippi/sitemcp logo

ryoppippi/sitemcp

Free

抓取整个网站并将其作为 MCP 服务器使用。

FreeFree tier
Inputs: urlOutputs: text
Type
Open Source
Visit Website GitHub (40 stars)

About ryoppippi/sitemcp

SiteMCP is an open-source tool that fetches an entire website and serves it as an MCP (Model Context Protocol) server, allowing AI assistants like Claude Code to access website documentation locally. Forked from egoist/sitefetch, it offers easy one-off usage via bunx, npx, or pnpx, and supports global installation. Features include configurable concurrency (e.g., --concurrency 10), tool name strategies (-t domain|subdomain|pathname) to customize MCP server names, and a maximum content length limit (-l, default 2000 characters). Created on April 8, 2025, during the early days of MCP, the project has since been archived by its author, who now recommends publishing documentation as static Markdown files or npm packages instead of using MCP for this purpose. The repository is read-only but remains available for reference.

Key Features

Fetches an entire website and serves it as an MCP server for AI assistants
One-off usage via bunx, npx, or pnpx (e.g., `bunx sitemcp https://example.com`)
Global installation with bun, npm, or pnpm
Configurable concurrency with `--concurrency` flag
Customizable tool name strategy: domain, subdomain, or pathname via `-t` flag
Max content length limit with `-l` flag (default 2000 characters)
Open-source (MIT license implied by GitHub) and free

Pros & Cons

Pros
  • Quick and easy setup with npm/bun/pnpm
  • Supports concurrent fetching to speed up large sites
  • Flexible tool naming adapts to different site structures
  • Configurable content length to avoid overwhelming AI context
  • Fully open-source with active history (now archived but usable)
Cons
  • Archived and no longer maintained; may have unresolved bugs or security issues
  • Overhead of serving static documentation via MCP is now considered an anti-pattern by the author
  • Depends on network availability and site structure for reliable fetching
  • Limited to text-based content; images and interactive elements are not served

Best For

Providing documentation from a website to an AI assistant (e.g., Claude Code Desktop) via MCPServing static website content locally for AI contextTesting MCP with real website data for development or experimentation

FAQ

What is SiteMCP?
SiteMCP is an open-source tool that fetches an entire website and serves it as an MCP (Model Context Protocol) server, enabling AI assistants to access the site's documentation locally.
How do I install SiteMCP?
You can use it without installation via `bunx sitemcp`, `npx sitemcp`, or `pnpx sitemcp`. For global installation, run `bun i -g sitemcp`, `npm i -g sitemcp`, or `pnpm i -g sitemcp`.
How do I use SiteMCP?
Run `sitemcp https://example.com` to fetch a site. You can add `--concurrency 10` for faster fetching, `-t domain|subdomain|pathname` to set the tool name strategy, and `-l 5000` to set max content length per page.
Can I customize the MCP server name?
Yes, use the `-t` flag with options: `domain` (default), `subdomain`, or `pathname`. For example, `sitemcp https://vite.dev -t domain` generates tool names like `indexOfVite` and `getDocumentOfVite`.
Why is SiteMCP archived?
The author archived the project because using MCP to serve static documentation is now considered an anti-pattern. They recommend publishing docs as static Markdown files or npm packages instead, as detailed in their blog posts.
Is SiteMCP still usable?
Yes, the source code is still available on GitHub and can be used, but it is no longer maintained. The repository is read-only and may not receive updates or fixes.