n8n-nodes-firecrawl-v2

Community node for Firecrawl v2 API on n8n. Scrape, crawl, map, search, and extract web content with full JavaScript rendering and AI-powered extraction.

Works against both Firecrawl Cloud and a self-hosted Firecrawl instance: point the credential Base URL wherever your API lives.

Requires self-hosted n8n. This node is not n8n-verified, and unverified community nodes are not available on n8n Cloud. If you are on n8n Cloud, our n8n-nodes-zalo-platform is verified and installs straight from the nodes panel.

Built by hecigo — Middleware & Integration Lab.

Guides: Firecrawl on n8n: picking the right operation (English) · Detailed Guide (Vietnamese)

Installation

Self-hosted n8n

1. Go to Settings > Community Nodes
2. Select Install
3. Enter n8n-nodes-firecrawl-v2
4. Agree to the risks and click Install

Manual (Self-Hosted)

cd ~/.n8n
npm install n8n-nodes-firecrawl-v2

Restart n8n after installation.

Credentials

| Field | Default | Description | |-------|---------|-------------| | Base URL | https://api.firecrawl.dev/v2 | Change for self-hosted. Must include /v2. | | API Key | | Your Firecrawl API key |

Authentication: Authorization: Bearer {apiKey}. Tested via POST /scrape on https://example.com.

Operations

1. Scrape

Scrape content from a single URL with JS rendering.

Endpoint: POST /scrape

| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | url | String | | Target URL (required) |

Scrape Options:

| Parameter | Default | Description | |-----------|---------|-------------| | formats | markdown | markdown, html, rawHtml, links, screenshot, json, summary, images, audio, changeTracking | | onlyMainContent | true | Strip headers, nav, footers | | includeTags | | CSS selectors to keep (e.g., article, .content) | | excludeTags | | CSS selectors to remove (e.g., nav, .sidebar) | | waitFor | 0 | Wait for JS render (ms). Increase for SPA pages. | | timeout | 30000 | Request timeout (ms), max 300,000 | | mobile | false | Emulate mobile viewport | | blockAds | true | Block ads and cookie popups | | proxy | auto | Proxy: auto, basic, enhanced | | locationCountry | | ISO country code (e.g., VN, US) | | locationLanguages | | Locale codes (e.g., vi-VN, en-US) |

Sample output:

{
  "markdown": "# Page Title\n\nExtracted content...",
  "metadata": {
    "title": "Page Title",
    "sourceURL": "https://example.com",
    "statusCode": 200
  }
}

2. Crawl

Crawl an entire website. Async job with optional polling.

Endpoint: POST /crawl

| Parameter | Default | Description | |-----------|---------|-------------| | crawlUrl | | Starting URL (required) | | waitForCompletion | false | Poll until job finishes | | maxPollTime | 300 | Max wait in seconds |

Crawl Options:

| Parameter | Default | Description | |-----------|---------|-------------| | limit | 50 | Max pages. Firecrawl's own default is 10000 if you omit the option. | | maxDiscoveryDepth | 2 | Max link depth | | includePaths | | Regex patterns to include (e.g., /blog/*) | | excludePaths | | Regex patterns to exclude (e.g., /admin/*) | | sitemap | include | include, skip, or only | | crawlEntireDomain | false | Follow sibling/parent links | | allowExternalLinks | false | Follow external links | | allowSubdomains | false | Crawl subdomains | | delay | 0 | Seconds between requests (forces concurrency=1) | | formats | markdown | Output format per page | | onlyMainContent | true | Strip boilerplate |

When waitForCompletion is off, output only contains the job id. Use Get Crawl Status to fetch results. Polling interval: 2 seconds.

3. Get Crawl Status

Endpoint: GET /crawl/{crawlId} | Parameter: crawlId (job ID)

4. Cancel Crawl

Endpoint: DELETE /crawl/{crawlId} | Parameter: cancelCrawlId (job ID)

5. Map

Discover all URLs on a website without scraping content. Faster than Crawl.

Endpoint: POST /map

| Parameter | Default | Description | |-----------|---------|-------------| | mapUrl | | Starting URL (required) | | search | | Search query to rank by relevance | | includeSubdomains | true | Include subdomain URLs | | limit | 50 | Max URLs, capped at 100,000 | | ignoreQueryParameters | true | Deduplicate by stripping query strings | | ignoreCache | false | Bypass sitemap cache |

6. Search

Web search with optional page scraping.

Endpoint: POST /search

| Parameter | Default | Description | |-----------|---------|-------------| | searchQuery | | Keywords, max 500 chars (required) | | limit | 50 | Results count, capped at 100 | | country | US | ISO country code | | tbs | Any Time | Time filter: past hour/day/week/month/year | | formats | markdown | Content format for results | | onlyMainContent | true | Strip boilerplate |

7. Extract

AI-powered structured data extraction using natural language prompts.

Endpoint: POST /extract

| Parameter | Default | Description | |-----------|---------|-------------| | extractUrls | | Comma-separated URLs (glob patterns supported: https://example.com/*) | | extractPrompt | | Natural language instruction | | extractSchema | | Optional JSON Schema for output structure | | extractWaitForCompletion | true | Wait for results (defaults ON, unlike Crawl/Batch) | | extractMaxPollTime | 300 | Max wait in seconds |

Extract Options:

| Parameter | Default | Description | |-----------|---------|-------------| | enableWebSearch | false | Use web search for additional data | | showSources | false | Include source URLs |

Example:

Prompt: "Extract company name, phone, address from this page"
Schema: {
  "type": "object",
  "properties": {
    "company_name": { "type": "string" },
    "phone": { "type": "string" },
    "address": { "type": "string" }
  }
}

8. Get Extract Status

Endpoint: GET /extract/{extractId} | Parameter: extractId (job ID)

9. Batch Scrape

Scrape multiple URLs asynchronously.

Endpoint: POST /batch/scrape

| Parameter | Default | Description | |-----------|---------|-------------| | batchUrls | | Comma-separated URLs | | batchWaitForCompletion | false | Wait for all to finish | | batchMaxPollTime | 300 | Max wait in seconds |

Options: formats, onlyMainContent, maxConcurrency.

10. Get Batch Scrape Status

Endpoint: GET /batch/scrape/{batchScrapeId} | Parameter: batchScrapeId (job ID)

Technical Notes

• Async operations (Crawl, Extract, Batch Scrape) return a job ID by default. Enable waitForCompletion to get results inline. Polling interval: 2 seconds.
• Extract defaults to waitForCompletion: true, while Crawl and Batch Scrape default to false.
• Scrape supports 10 formats (including json, summary, audio). Crawl, Search, and Batch Scrape support 5 basic formats.
• Comma-separated inputs: includeTags, excludeTags, includePaths, excludePaths, extractUrls, batchUrls all accept comma-separated lists.
• Self-hosted Base URL must include /v2 (e.g., http://firecrawl:3002/v2).
• Error handling: Supports continueOnFail. On error, output is { "error": "message" }.

Workflow Examples

Competitive intelligence:

Schedule Trigger (weekly)
  -> Firecrawl: Map (competitor URL)
  -> Firecrawl: Batch Scrape (URLs from Map)
  -> Code Node (diff with last week)
  -> Google Sheets + Slack notification

AI data extraction:

Manual Trigger
  -> Firecrawl: Extract (directory URL, prompt, schema)
  -> Google Sheets: Append rows

Content monitoring:

Schedule Trigger (daily)
  -> Firecrawl: Scrape (formats: changeTracking)
  -> IF (changes detected) -> Email alert

Compatibility

• n8n: >= 1.83. The node imports NodeConnectionTypes from n8n-workflow, which first shipped in [email protected]. On older n8n the package fails to load. Stay on 1.0.17 if you cannot upgrade n8n.
• Firecrawl API: v2
• Tested with self-hosted Firecrawl and Firecrawl Cloud

About

hecigo is a middleware and integration lab. We build and run the layer that makes disconnected business systems talk to each other — with the right data, and nothing lost. Need custom n8n nodes, self-hosted Firecrawl deployment, or scraping workflow design? Get in touch.

Other n8n nodes by hecigo: n8n-nodes-zalo-platform — Zalo Bot Platform on n8n: webhook trigger, replies, photos, stickers and chat actions. n8n-verified, runs on n8n Cloud.

Consulting partner: THE NEXOVA — Bridge Strategy to Execution.

License

MIT