Firecrawl CLI skills
Search the web, scrape pages to clean markdown, crawl sites and watch pages for changes from inside an agent session.
Install
-
CLI (pinned)
npm install -g firecrawl-cli@1.24.4Pin the version. The vendor's one-liner uses @latest and installs into every agent it detects.
-
Sign in
firecrawl login --browser -
Check it works
firecrawl --status
Thirteen skills that teach an agent to drive the Firecrawl CLI: when to search versus scrape, when a page needs a real browser, and how to recover when a job fails.
When to use it
- An agent needs current web content — not what the model remembers.
- You want pages as clean markdown to cite or store.
- You need to watch a page and get alerted when it changes (
firecrawl-monitor).
Before you install
Firecrawl’s quick-start runs npx firecrawl-cli@latest init --all, which installs skills into
every coding agent on the machine without asking. Pin the CLI version instead, and rescan
after upgrading. Scrapes save into .firecrawl/ by default — add it to .gitignore.
- Publisher
- Firecrawl
- Source
- github.com/firecrawl/cli
- Works with
- Claude Code, Codex, Cursor, Gemini
- Categories
- Research & web data, MCP & integrations
- License
- ISC
- Cost
- Free tier, paid plans