Firecrawl Module
Read the web from a workflow through Firecrawl: a page as clean markdown, the list of a website's pages, web search results, or structured data extracted from a page with an AI prompt.
Typical uses: research a prospect's website before first contact, keep a product or price list in sync with a supplier's site, read a page a customer sent in chat, or feed a page to an AI task.
Requirements
- A Firecrawl account and API key. Every call uses Firecrawl credits on that account: Extract data and Read each result page cost more than a plain read.
Configuration
| Setting | Required | Notes |
|---|---|---|
| Firecrawl API key | yes | From the Firecrawl dashboard → API Keys. Stored securely. |
| API address | no | Only for a self-hosted Firecrawl; must be https. Empty: Firecrawl cloud. |
| Default timeout (seconds) | no | 5–300. Empty: 60. A task can set its own. |
Firecrawl
One task with four operations. The result is written to the Output variable.
Read page
Reads one page, JavaScript rendered, as markdown.
| Property | Notes |
|---|---|
| URL | The page. www.example.it is read as https://www.example.it. |
| Main content only | On by default: leaves out menus, header and footer. Turn it off to keep contacts in the footer. |
| Include links | Also return every link on the page. |
Output: {page} with Url, FinalUrl, Title, Description, Language, StatusCode, Markdown, Truncated and, with links on, Links.
List site pages
Lists a website's pages from its sitemap and links.
| Property | Notes |
|---|---|
| URL | The site. |
| Order by relevance to | Optional words, e.g. contatti servizi resi: the best matches come first. |
| Maximum results | 1–5000. Empty: 50. |
Output: {pages} with Url, Count and Pages (each Url, Title, Description).
Search the web
| Property | Notes |
|---|---|
| Search query | Up to 500 characters; search operators such as site: work. |
| Maximum results | 1–100. Empty: 10. |
| Read each result page | Also return each result's markdown (one extra credit per result). |
Output: {found} with Query, Count and Results (each Url, Title, Description, Markdown).
Extract data
Reads a page and returns the fields you describe, as JSON.
| Property | Notes |
|---|---|
| URL | The page. |
| What to extract | Plain-words description of the fields, e.g. the services offered and every contact email. |
| JSON schema | Optional; use it to get the same fields every time. |
Output: the extracted object, e.g. {facts.services}.
Advanced options
| Property | Notes |
|---|---|
| Country | Two-letter code (e.g. IT) the search or visit appears to come from. |
| Maximum characters | Markdown longer than this is cut (Read page, Search). Empty: 50000. |
| Timeout (seconds) | 5–300. Empty: the module's default. |
Errors
The task takes its error flow when the input is invalid, the API key is refused (401/403), the account has no credits (402), the rate limit is reached (429), the site blocks automated access, or Firecrawl times out. The log says which.
Respect each website's terms of use; the module never tries to get around a block, a login or a captcha.