Setup & Installation
What This Skill Does
Fetches one previously-stored page from Crawlbase Cloud Storage by its request id or original URL, without re-crawling. The stored body comes back as a full JSON record, raw HTML, or converted Markdown. Storage is split per token, so JS-rendered pages and screenshots live in a separate silo addressed with use_js_token.
Crawl once, read many times: stored pages come back instantly and free of charge, so agents can revisit past crawls without hitting the live site again.
When to use it
- Re-reading a crawled page later without spending another crawl request
- Converting a stored HTML page to Markdown for summarization
- Retrieving a JS-rendered page stored under the JS token silo
- Looking up a stored page by its original URL when the rid is unknown
More from Crawlbase
The MCP server behind these skills (npm @crawlbase/mcp). Connects AI agents and LLMs to real-time web data with JavaScript rendering, automatic proxy rotation, and anti-bot protection. Works with Claude, Cursor, and Windsurf.
LangChain document loader, tool, and retriever backed by the Crawlbase Crawling API. Returns clean Markdown documents ready for chunking (pip install langchain-crawlbase).
Native Crawlbase node for n8n. Adds Crawling API credentials and options like JS rendering, geo-targeting, screenshots, and storage to your workflows without hand-wiring HTTP Request nodes.