Skip to main content

linkinator-mcp

A Model Context Protocol (MCP) server that enables AI assistants to perform link validation and broken link checking.

linkinator-mcp

npm version MIT License

The linkinator MCP server integrates the power of linkinator with AI assistants through the Model Context Protocol. This enables Claude and other AI tools to scan webpages and local files for broken links directly from chat conversations.

Features​

  • 🔍 Comprehensive Link Validation - Scan websites and local files for broken or inaccessible links
  • 🌐 Recursive Crawling - Follow links within the same domain automatically
  • 📝 Multi-Format Support - Check links in HTML, CSS, and Markdown files
  • ⚓ Fragment Validation - Verify URL anchors and internal page references
  • ⚙️ Extensive Configuration - Customize timeouts, retry logic, SSL handling, and more
  • 📊 Detailed Reporting - Get comprehensive results with HTTP status codes and error categorization

Installation​

The easiest way to install is using the install-mcp tool:

npx install-mcp linkinator-mcp --client claude

This will automatically configure the MCP server for Claude Desktop, Claude Code, Cursor, Cline, and other MCP clients.

Manual Installation​

If you prefer manual configuration, you can add the server to your MCP client's config file.

Claude Desktop (macOS/Linux):

~/Library/Application Support/Claude/claude_desktop_config.json

Claude Desktop (Windows):

%APPDATA%\Claude\claude_desktop_config.json

Claude Code (macOS/Linux):

~/.config/claude-code/config.json

Claude Code (Windows):

%APPDATA%\claude-code\config.json

Add the following to your config:

{
  "mcpServers": {
    "linkinator": {
      "command": "npx",
      "args": ["linkinator-mcp"]
    }
  }
}

For a global installation:

npm install -g linkinator-mcp

Then use:

{
  "mcpServers": {
    "linkinator": {
      "command": "linkinator-mcp"
    }
  }
}

Usage​

Once installed, you can ask your AI assistant to check links:

  • "Check all the links on https://example.com"
  • "Scan this website recursively for broken links"
  • "Validate the links in my local documentation folder"
  • "Check for broken anchors and fragments on this page"

Configuration Options​

The scan_page tool supports extensive configuration:

Essential​

  • path - URL or local file path to scan (required)

Connection Settings​

  • concurrency - Number of simultaneous connections
  • port - Port for local file server
  • timeout - Request timeout in milliseconds

Crawling Behavior​

  • recurse - Follow same-domain links recursively
  • serverRoot - Custom server root for local scans
  • cleanUrls - Support extensionless URLs

Content Parsing​

  • markdown - Parse and scan markdown files
  • checkCss - Extract and validate URLs in CSS
  • checkFragments - Verify anchor identifiers

Filtering​

  • linksToSkip - Regex patterns for URLs to skip
  • userAgent - Custom user agent string

Retry Logic​

  • retry - Enable automatic retries for HTTP 429
  • retryErrors - Retry on 5xx errors
  • retryErrorsCount - Maximum retry attempts
  • retryErrorsJitter - Random delay between retries

Security​

  • allowInsecureCerts - Accept invalid SSL certificates

Output​

Results include:

  • Total links scanned
  • Successful connections
  • Broken links categorized by HTTP status code
  • Source pages for each broken link
  • Detailed error information

Example Results​

Scan Complete!
✓ Total links: 127
✓ Successful: 125
✗ Broken: 2

Broken Links:
  404 Not Found
    - https://example.com/missing-page
      → Found on: https://example.com/index.html

License​

MIT