Firecrawl
Arcade.dev LLM tools for reading the web via Firecrawl
4.0.0Firecrawl is a web-reading service; this toolkit gives Arcade tools access to its crawling, scraping, searching, and structured-extraction capabilities so LLMs can retrieve and process web content.
Capabilities
- Single-page and multi-page retrieval: Scrape a known URL for its full text, or crawl an entire site and collect content across many pages — with async job management (status checks, partial data reads, cancellation) for long-running crawls.
- Site mapping: Enumerate all URLs on a website without fetching page content, useful for planning targeted scrapes or crawls.
- Web and specialized search: Search the open web (with optional inline page reading), developer documentation and repositories, GitHub issues, and scientific literature (PubMed, bioRxiv, medRxiv, arXiv) through purpose-built search tools.
- Structured data extraction: Describe the fields you want in plain language (products, prices, contacts, etc.) and extract structured records from one or more pages, with async status polling for long-running extractions.
Secrets
FIRECRAWL_API_KEY — A Firecrawl API key that authenticates all requests to the Firecrawl API. Obtain it by signing in to the Firecrawl dashboard, navigating to the API Keys section, and creating a new key. The free tier provides a limited number of credits; paid plans unlock higher limits and additional features. Copy the key immediately after creation — it is not shown again in full.
Configure secrets in Arcade at https://docs.arcade.dev/en/guides/create-tools/tool-basics/create-tool-secrets or directly via the dashboard at https://api.arcade.dev/dashboard/auth/secrets.
Available tools(12)
| Tool name | Description | Secrets | |
|---|---|---|---|
Stop a crawl that is still running.
Pages the crawl already collected stay available through GetCrawlData. | 1 | ||
Read many pages of one website and return each page's content.
A crawl that outruns wait_seconds keeps running: the result carries its
job_id, and GetCrawlStatus, GetCrawlData, and CancelCrawl take it from there. | 1 | ||
Collect structured records from the web from a plain-language description.
Use this when the answer is fields such as products, prices, or contacts.
Use ScrapeUrl or CrawlWebsite when the answer is the page text itself. An
extraction that outruns wait_seconds keeps running, and GetExtractStatus
takes its job_id. | 1 | ||
Read the pages a crawl has collected so far. | 1 | ||
Check how far a crawl has progressed, without fetching its pages. | 1 | ||
Collect the results of an extraction that is already running. | 1 | ||
List the URLs on a website, without reading the pages.
Use CrawlWebsite to read the content of many pages, or ScrapeUrl for one. | 1 | ||
Read one web page whose URL is already known and return its content.
Use Search instead when the URL is unknown, MapWebsite to list a site's
URLs, CrawlWebsite to read many pages of one site, and ExtractData to pull
structured fields rather than page text. | 1 | ||
Search the web and optionally read each result page in the same call.
Use this when no URL is known yet. Use SearchResearchPapers for scientific
literature and SearchDeveloperDocs for API and code documentation. | 1 | ||
Search developer documentation and repositories, quoting the text that matched.
Use this to answer a question about how a library or API works. Use Search
for the open web, and SearchGithubIssues to look only at issues. | 1 | ||
Search GitHub issues across indexed repositories.
Use SearchDeveloperDocs instead to search documentation, READMEs, and pull
requests alongside issues. | 1 | ||
Search published scientific papers and return their abstracts.
This reads a corpus of paper records drawn from PubMed, bioRxiv, medRxiv,
and arXiv. To find ordinary web pages that happen to sit on academic sites,
use Search narrowed to research instead. | 1 |