mcp-read-website-fast
MCP server that reads sites on your machine and hands the agent short Markdown with links, keeping raw HTML out of context.
- 163 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The server visits any address the agent names and stores a page cache on disk.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/mcp-read-website-fastDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The server visits any address the agent names and stores a page cache on disk.
This is third-party code. Review the repository files before installing.
What it does
mcp-read-website-fast fetches a page locally, strips navigation and noise with Mozilla Readability (the engine behind Firefox Reader View) and returns Markdown with links preserved. The model never has to parse raw HTML, so replies are shorter and cheaper in tokens. Its single tool, `read_website`, takes a `url` and an optional `pages` count: how many pages to crawl (1 by default, 100 at most). The server respects `robots.txt`, rate-limits requests and caches responses on disk under a SHA-256 hash of the address.
Who it is for. Developers whose agent reads documentation and articles from links and loses context to extra HTML.
Good fit when
- The agent reads library documentation from a link and needs only the page text.
- A long article has to become Markdown that keeps its source links.
- Several related pages of a site are crawled in one call with the `pages` parameter.
- Site material is prepared for RAG or a knowledge base and split into chunks.
Not a fit when
- The page is rendered by JavaScript, which the server does not process.
- You need clicks, forms or account sign-in: the server only downloads and converts a page.
- A site blocks sections with `robots.txt` and the blocked section is the one you need.
Example request
Read https://example.com/docs/setup and summarize the install steps, keeping the links to each section.MCP
- Transport
- stdio
- Authentication
- not required
Limitations
Pages that need JavaScript to render are not read; the troubleshooting section says so directly. Some sites block automated requests; a custom user agent helps in command mode but success is not guaranteed. Pages behind a login are out of reach because the server goes online without your session. The disk cache grows with crawling and is cleared separately through the `read-website-fast://clear-cache` resource. The package is still at 0.x, so behavior may change between releases. Node.js is required: the package runs through `npx`.
How to disable. Remove the connection with `claude mcp remove read-website-fast` or delete the entry from the client's `mcp.json`. Clear the page cache separately.
FAQ
How do I connect read-website-fast to Claude Code?
Run `claude mcp add read-website-fast -s user -- npx -y @just-every/mcp-read-website-fast`. Cursor has a deeplink, VS Code uses `code --add-mcp`, and other clients take a JSON entry with `npx` under `mcpServers`.
Which tools does the server expose?
The `read_website` tool and two resources. `read-website-fast://status` shows cache statistics and `read-website-fast://clear-cache` empties it.
How many pages does one call crawl?
One by default. The `pages` parameter raises the limit to 100.
Does the server read pages rendered with JavaScript?
No. It downloads HTML and converts it without launching a browser. Such sites need a tool with a real browser.
Does the server follow robots.txt?
Yes, by default. The `--no-robots` flag exists only in the development command mode.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server that opens pages in headless Chromium through Playwright and hands the agent the main text as Markdown, even from JavaScript-heavy sites
A Rust server that turns pages into Markdown or JSON: search, scrape, site map, crawl and PDF through MCP, a CLI and a Firecrawl-compatible API
A plugin that crawls sites and turns pages into Markdown with metadata: seven skills and an MCP server with nine tools
A self-hosted Rust service for agents: pages, PDFs and Office files to Markdown, Firecrawl-style API, MCP included
Browserless.io MCP server: a cloud browser for the agent with page reads, site crawls, Lighthouse audits and click-driven sessions
A web search MCP server that gives the agent full page text at once: Stack Overflow answers, GitHub threads, arXiv papers
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| mcp-read-website-fast (this card) | no | not required | 163 | |
| Fetcher MCP | no | not required | 1,088 | opens pages in headless Chromium and reads JavaScript sites; mcp-read-website-fast downloads HTML quickly without a browser |
| Firecrawl MCP Server | yes | API key | 7,765 | runs through the Firecrawl cloud API with a key; mcp-read-website-fast works locally and needs no key |
README in short
The crawling and conversion core lives in the `@just-every/crawl` package, and the server adds the MCP layer. After a crash the process restarts itself, at most 10 times a minute, and shuts down cleanly on SIGINT and SIGTERM.