webfetch

name: webfetch description: Fetch and read live content from URLs, including markdown conversion for HTML pages.

Web Fetch Skill

When to use this skill

Use this skill when a user asks for current or exact content from a URL, such as:

• "Summarize this page"
• "What does this doc say?"
• "Fetch this link and extract the key points"

Tool to use

Call the webFetch tool with:

• url (required): must start with http:// or https://
• format (optional): markdown (default), text, or html
• timeout (optional): seconds, max 120

Usage guidelines

1. Prefer format: "markdown" for article/docs pages.
2. Use format: "text" when the user wants plain extracted text.
3. Use format: "html" only when raw HTML is explicitly requested.
4. If the tool returns an error, explain it clearly and offer a retry suggestion.
5. If successful, provide a concise summary first, then include specific details the user asked for.

Output handling

• The tool response includes a title and output.
• For image URLs, it may return imageDataUrl; describe what happened instead of dumping the entire data URL unless explicitly requested.