Web Fetch

Web Fetch

Fetch any URL and return its content as readable text. HTML is automatically stripped for clarity.

When to Use

• Retrieving the full content of a web page by URL
• Reading documentation, articles, or API responses from a known URL
• Fetching raw JSON or text content from an API endpoint
• Following up on URLs found via web_search

When NOT to Use

• Searching the web for information — use the web_search tool instead
• Fetching content that requires authentication (login-gated pages)
• Downloading binary files (images, PDFs, archives)

Tools

web_fetch

Fetch a URL and return its text content. HTML tags are stripped for readability.

Parameters:

Returns: { "result": "..." } — plain text content of the page with HTML stripped:

Page title and heading text
Body content, paragraphs, code examples...
[truncated, N chars omitted]

JSON responses are returned as-is without stripping. On failure, returns { "error": "..." }.

Examples:

• { "url": "https://docs.example.com/guide", "maxChars": 5000 } — fetch a docs page
• { "url": "https://api.example.com/data.json", "extractMode": "raw" } — fetch raw JSON

Notes

• Uses Node.js fetch with a 30-second timeout
• Automatically strips <script> and <style> tags and all HTML markup
• JSON responses are detected and returned without stripping
• Output is truncated at maxChars with a note indicating omitted content
• Blocks requests to private/internal IP addresses for security