Firecrawl
Connect a Firecrawl account on every action. Outputs depend on toggles (links, screenshots, etc.).Shared scrape options
Scrape URL and Crawl Website reuse the same scrape configuration after the URL fields:| Input | Purpose |
|---|---|
| Format | Markdown, HTML, or Raw HTML for main content. |
| Include links | Adds a links output: list of URLs found on the page. |
| Include screenshot | Adds a screenshot output (image file). When on, Full page screenshot chooses viewport vs full page. |
| Only main content | Strips chrome (nav, footer, etc.); default on. |
| Mobile | Emulate mobile user agent (responsive layouts, mobile screenshots). |
| Headers | Extra request headers (cookies, User-Agent, etc.) as name/value rows. |
| Wait for | Milliseconds to wait after load before scraping. |
| Remove base64 images | Strips huge inline images from output; keeps alt text with a placeholder (default on). |
- content — Page text in the selected format.
- links — If enabled, repeatable list of link URLs.
- screenshot — If enabled, image file.
Scrape URL
Inputs- URL to fetch.
- Firecrawl account.
- Shared scrape options above.
Crawl Website
Walks internal links starting from a base URL. Inputs- URL — Starting URL.
- Limit — Max pages (1–1000, default 100).
- Max depth — How “deep” path segments can go (1–100; effectively limits URL path depth).
- Ignore query parameters — Treat
?x=1variants as the same path for deduplication. - Allow backward links — Follow links to sibling/parent paths, not only deeper nested paths.
- Shared scrape options for each page.
- total — Count of pages crawled.
- pages — Repeatable: each item includes url, content, links (array), and screenshot (when enabled in scrape options).
Search Web
Runs a web search via Firecrawl. Inputs- Query
- Limit — Result count (1–100). Hidden when you wire the results output as a fixed-size list in the graph—the number of connected “slots” becomes the limit.
- Location — Optional geo bias string (e.g.
"Germany"). - Time filter — Any time, past hour/day/week/month/year.
- total — Number of results returned.
- results — Repeatable; each result exposes title, description (snippet), and url.