| name | firecrawl |
| description | Self-hosted Firecrawl — deep web crawling, scraping, and search for AI agents. Converts websites into clean markdown. |
| when_to_use | When you need to crawl entire websites, extract documentation, scrape JS-rendered pages, or get structured data from complex sites. NOT for single URL reads (use web_fetch) or searching (use web_search). |
| allowed_tools | ["exec","file_write"] |
| context | inline |
Firecrawl Plugin — Self-Hosted Web Crawler
What This Plugin Does
Firecrawl converts websites into clean, LLM-ready markdown. It handles JavaScript rendering, follows links, and extracts structured content.
Installation (Self-Hosted)
Option 1: Docker (Recommended)
docker run -d --name firecrawl \
-p 3002:3002 \
-e FIRECRAWL_PORT=3002 \
ghcr.io/firecrawl/firecrawl:latest
Option 2: Docker Compose (add to existing stack)
Add this to your docker-compose.yml:
firecrawl:
image: ghcr.io/firecrawl/firecrawl:latest
ports: ["3002:3002"]
environment:
- FIRECRAWL_PORT=3002
restart: unless-stopped
After Installation
Set the environment variable so Qorven uses your self-hosted instance:
export FIRECRAWL_URL=http://localhost:3002/v1
When to Use Which Tool
| Need | Tool | Cost |
|---|
| Search the web | web_search | Free (Tavily/DDG) |
| Read one URL | web_fetch | Free (built-in) |
| Deep crawl a site | firecrawl (this plugin) | Free if self-hosted |
| Custom CSS extraction | scrapling plugin | Free |
| JS-rendered pages | firecrawl scrape mode | Free if self-hosted |
API Usage (if using hosted API)
Get an API key from https://firecrawl.dev — set CRAWL4AI_API_TOKEN=fc-xxx
Capabilities
- Scrape: Single page → clean markdown (handles JS)
- Crawl: Follow links → multiple pages as markdown
- Search: Search + scrape in one call
- Extract: Structured data extraction with schemas
