Claude Code plugin
One command adds Better Fetch to Claude Code: nine skills and a scraper subagent that teach the agent to fetch the web well, plus the MCP connector wired up and ready.
Install
Run these in Claude Code. You'll be prompted for your API key once — it's stored securely and used to authenticate the connector.
/plugin marketplace add better-fetch/claude-plugins /plugin install better-fetch@better-fetch
No key yet? Create one. The plugin source is public on GitHub.
What you get
Skills auto-activate when relevant; the
better-fetch MCP tools are available to call directly.- extract-structured-data — End-to-end: find a page's data API and pull clean JSON, falling back to rendered HTML.
- fetch-page — Render JavaScript pages; sticky sessions, custom headers, country locale/timezone defaults, and fingerprint control.
- scrape-json-api — Pull JSON from APIs behind bot protection or geo-fences, with the headers they require.
- discover-apis — Capture a page's network calls — with response previews — to find its internal APIs.
- crawl-pages — Client-orchestrated multi-page retrieval: pagination, sitemaps, sticky sessions, and polite pacing.
- screenshot-page — Capture full-page or viewport screenshots.
- bypass-bot-walls — Classify block responses and escalate browser identity or routing deliberately; collect cf_clearance tokens when issued.
- webfetch-fallback — Escalate automatically when a plain fetch comes back blocked, empty, or geo-walled.
- usage — Check your plan, credits used, remaining quota, and stored browser session usage.
- web-scraper — agent: coordinates bounded fetches and extraction in its own context, then returns only the extracted data. This is not a hosted durable crawl queue.
Also in the marketplace: site QA
For your own site rather than scraping: better-fetch-site-qa adds
geo-check (check country-coherent browser identity), seo-render-check (what crawlers see vs users), and monitor-page-changes./plugin install better-fetch-site-qa@better-fetch