M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
M

MCP Read Website Fast

Community
Fast, token-efficient web content extraction for AI agents - converts websites to clean Markdown.
How to install ↓ Compare GitHub source repository ↗
Category
Browser #19 of 32
Stars
★ 161 Very popular
Transport
stdio (local process)
Runtime
Node.js
Credentials
No credential needed
License
MIT
Last commit
Tools
1
51FMRS · D

A lightweight, fast MCP server for web scraping and markdown conversion, but with limitations regarding JavaScript-dependent sites and potential cache management issues.

Strongest · Documentation 14/20 Weakest · Reliability 7/20

Reliability
7/20
Security and permissions
10/20
Maintenance
7/20
Documentation
14/20
Setup experience
13/20
Why each score
Reliability 7/20
Evidence: README describes the read_website tool, status/clear-cache resources, official MCP SDK with lazy loading and auto-restart; the happy path appears plausible. Deductions: no package.json, source, tests or CI workflows were supplied, and the server manifest is not cached, so the handshake, tool registration, cache/robots behavior and error handling cannot be verified; reliability is kept below 12 due to static calibration.
Security and permissions 10/20
Evidence: no credentials, tokens, or install examples containing real secrets were found, and default robots.txt indicates restrained crawling; no red line is triggered. Deductions: source code is unavailable to prove absence of covert exfiltration; arbitrary URL fetching is allowed, the clear-cache resource has a side effect of clearing data, and permission scoping/confirmation and data-flow disclosure are incomplete.
Maintenance 7/20
Evidence: the repository is not archived and has an MIT license; README shows npm and GitHub Actions Release badges. Deductions: no commit/release dates, dependency update records, issue response evidence, or security-response channel were supplied; badges, stars, and the single open issue alone do not demonstrate sustained maintenance.
Documentation 14/20
Evidence: README provides install examples for Claude Code/VS Code/Cursor/JetBrains/raw JSON, tool parameters, resources, CLI options, architecture, development, and troubleshooting. Deductions: the server manifest is absent, there is no formal schema or cost detail, the features list is duplicated, and some limitations appear only in troubleshooting.
Setup experience 13/20
Evidence: installation is essentially one-line npx commands with ready-to-paste config fragments for multiple clients; few steps and direct examples. Deductions: Node version/platform requirements are not stated, the npx version is unpinned, and no CI or tests serve as verifiable execution evidence; setup is capped at 15 by static calibration.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the network

Best for

  • Claude Code users
  • AI agents that frequently need to read web pages
  • Developers who want to minimize token usage

Not for

  • Websites that require JavaScript rendering
  • Sites with complex authentication requirements
  • Non-technical users who are not comfortable with CLI

Required permissions

  • Network access to fetch URLs
  • File system access for caching (uses .cache directory)
  • Ability to execute npx commands

Risks and side effects

  • Some websites may block automated access
  • Potential token consumption due to external dependencies
  • Cross-origin crawling (disabled by default) could violate website terms

Setup

Before you start

Runtime：Node.js

For Claude Code: claude mcp add read-website-fast -s user -- npx -y @just-every/mcp-read-website-fast. For other clients, copy the provided JSON into your mcp config.

claude_desktop_config.json 
{
  "mcpServers": {
    "read-website-fast": {
      "command": "npx",
      "args": [
        "-y",
        "@just-every/mcp-read-website-fast"
      ]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "read-website-fast": {
      "command": "npx",
      "args": [
        "-y",
        "@just-every/mcp-read-website-fast"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add read-website-fast -- npx -y @just-every/mcp-read-website-fast

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

After install, the read_website tool should appear in your client's tool list; ask it to read https://example.com and a clean Markdown response confirms the connection works.

Troubleshooting

  1. Cache issues: run `npm run dev clear-cache`
  2. Timeout errors: increase timeout with `-t`, check network connectivity
  3. Content not extracted: try custom user agent with `-u`, check if site requires JavaScript

Things to try

Once connected, you can ask your AI assistant things like:

  • Read https://example.com and convert it to Markdown for me
  • Fetch https://example.com and crawl up to 5 pages
  • Convert this documentation page to Markdown, keeping the links
  • Check the current cache statistics

Tools 1

read_website read-only
Fetches a webpage and converts it to clean markdown. Parameters: url (required), pages (optional, default 1, max 100).

Use cases

Quickly scrape web content for LLM processing
Build knowledge graphs by crawling multiple pages
Extract content from websites while respecting robots.txt

Supported clients

Claude Desktop
VS Code
Cursor
JetBrains IDEs

Listed from the project's documentation, not tested by this site.

Overview

This MCP server fetches web pages locally, strips noise, and converts content to clean Markdown while preserving links. It features fast startup, smart caching, polite crawling, and configurable depth crawling, designed for minimal token usage.

Similar servers

Compare with other MCP servers

Source revision 78d0bf83a904 Data synced 2026-10-12 Read the FMRS scoring method