Firecrawl
Search, scrape, map, and crawl public HTTPS websites
Firecrawl turns public web pages into content a project can search or process. A project owner supplies the Firecrawl API key when the AI requests the connector. No administrator OAuth credentials are required; administrators control whether the connector is visible.
Available operations
| Operation | What it does | Limit per minute |
|---|---|---|
scrape | Scrapes one HTTPS page into selected formats | 20 |
search | Searches the web | 20 |
map_site | Discovers URLs on one site | 10 |
start_crawl | Starts a crawl of up to 1,000 pages and depth 20 | 5 |
get_crawl_status | Reads crawl status and available results | 30 |
The connector accepts public HTTPS URLs. It rejects HTTP and credential-bearing URLs. Site access rules, Firecrawl plan limits, and provider billing still apply.
Connect and verify Firecrawl
Ask the AI for a concrete source, such as: “Index the public help pages under https://example.com/help.” Enter the API key in the secure connector form; do not paste it into chat or save it as a project secret.
- Scrape one public page and confirm its title or a known sentence.
- Map the same site and confirm expected URLs.
- Start a crawl with a small limit, then poll
get_crawl_statusuntil Firecrawl reports completion. - Compare one returned page with the live source.
If a page is empty, check the site's robots rules and whether it needs browser rendering or authentication. For private data, use a connector designed for that system instead of placing credentials in a URL.