WebFetch pulls content from any URL into the conversation - docs, specs, issue threads, API references.

What it does

Given a URL, WebFetch downloads the page, extracts the main content, and returns it as text. It handles JavaScript-rendered pages, so SPAs and docs sites with client-side routing work correctly. You can point Claude at an official doc page and have it answer questions grounded in real content instead of stale training data.

When to use it

Pulling in docs for a library Claude might not know well.
Reading an issue or RFC before making a change.
Quoting from a spec or standard in a PR description.
Fact-checking before writing code that depends on external behavior.

Gotchas

WebFetch is not a general-purpose scraper. Heavy sites or authenticated pages may fail or return partial content.
Rate limits apply. Batching twenty fetches in a row can get throttled.
Content is truncated for very long pages. Ask for a specific section if you can.

Official docs: https://code.claude.com/docs/en/tools-reference.md