Firecrawl API guide
Firecrawl Space is an independent guide. Use official Firecrawl docs as the canonical implementation source before shipping production code.
| Endpoint | Use when | Output fit | Production checks |
|---|---|---|---|
| Search | The job starts with a query and needs sources. | Search results plus page content for agent context. | Log query, source URLs, ranking assumptions, and freshness. |
| Scrape | You know the URL and need clean content. | Markdown, HTML, screenshot, metadata, or structured JSON. | Validate allowed URLs, empty states, redirects, and schema drift. |
| Crawl | You need many pages from one site. | Async job with page-level documents. | Set limits, respect robots and policies, monitor job status, dedupe URLs. |
| Map | You need URL discovery before extraction. | Candidate links with titles and descriptions where available. | Filter admin, account, duplicate, and low-value paths before crawling. |
| Batch scrape | You already have a URL list. | Async extraction over known targets. | Track per-URL status, retry policy, and partial failure handling. |
| Agent | The task is research-like and does not start with a fixed URL. | Natural-language result with sources, optionally schema-shaped. | Keep source citations, model choice, prompt, and audit notes. |
| Interact | A page requires clicking, typing, scrolling, or waiting before extraction. | Post-action scrape result and live interaction context. | Avoid private areas, consent barriers, and fragile selectors. |
import { Firecrawl } from 'firecrawl';
const app = new Firecrawl({ apiKey: process.env.FIRECRAWL_API_KEY });
const doc = await app.scrape('https://example.com', {
formats: ['markdown'],
});
console.log(doc.metadata?.sourceURL, doc.markdown?.slice(0, 500));