Web Fetch


ID	`web_fetch`
Category	Network
Features	None
Dependencies	None

Fetch content from URLs and convert HTML to markdown or plain text. Powered by fetchkit with built-in SSRF protection.

Tools

`web_fetch`

Fetch a URL and return its content.

Parameter	Type	Required	Description
`url`	string	yes	URL to fetch
`method`	string	no	HTTP method (default: `GET`)
`as_markdown`	boolean	no	Convert HTML to markdown
`as_text`	boolean	no	Convert HTML to plain text

Returns: content body, status code, metadata (size, content type, filename, last modified).

Use Cases

Research — fetch documentation, API references, or web articles
Data collection — retrieve JSON/CSV data from public APIs
Content extraction — convert web pages to clean markdown for processing
Link verification — check if URLs are reachable and inspect response metadata

Example

User: Summarize the Python 3.12 changelog

Agent:
  → web_fetch({ url: "https://docs.python.org/3/whatsnew/3.12.html", as_markdown: true })
  ← (markdown content of the changelog page)

  "Python 3.12 introduces several key changes: ..."

Notes

Timeouts: 1s for first byte, 30s for body. Partial content returned on body timeout.
Binary content: Images, PDFs, etc. return metadata only (content type, size) — not the binary data.
SSRF protection: Private IPs (loopback, RFC1918, link-local, CGNAT) are blocked by default with DNS pinning to prevent rebinding attacks.
Excessive newlines: Automatically filtered from converted content.