firecrawl/cli

🔥 Firecrawl CLI – What It Is

The Firecrawl CLI is a command‑line tool that lets you interact with the Firecrawl platform directly from your terminal. Firecrawl is an AI‑powered web‑scraping, search, and research service that can:

• Crawl and scrape webpages (including JavaScript‑rendered sites)
• Perform web‑searches with highlighted snippets
• Query a specialised developer‑code index and a massive research‑paper index
• Run higher‑level “agent jobs” such as SEO audits, lead‑list generation, or knowledge‑base creation.

The CLI bundles all of these capabilities into simple commands (scrape, search, developer, research, etc.) and can also install skills that teach supported AI coding agents (Claude Code, Cursor, OpenHands, etc.) how to use Firecrawl automatically.

---

Quick‑Start & Installation

# Global install via npm
npm install -g firecrawl-cli

# One‑liner that installs the CLI, logs you in, and registers skills for any detected AI coding editors
npx -y firecrawl-cli@latest init -y --browser

• -y – runs the setup non‑interactively.
• --browser – opens a browser window for OAuth‑style login.
• Skills are automatically added to every detected “harness” (Claude Code, Cursor, etc.) unless you scope them with --agent.

---

Core Commands

Command	What It Does	Typical Options
scrape <url>	Pulls content from a page. Returns markdown by default but can also output raw HTML, JSON, images, screenshots, or an AI‑generated summary.	--format <list>, --html, --summary, --only-main-content, --screenshot, --output <file>
search "query"	Runs a web search (Google‑style) with optional highlights. Can also scrape the returned pages in one step.	--limit, `--sources web
developer "query"	Searches the Developer Index – a curated collection of public repos, GitHub issues, PRs, READMEs, and docs. Ideal for programming‑question answering.	--limit, --json, --pretty
research …	Queries the Research Paper Index (≈43 M abstracts, mostly biomedical plus arXiv). Supports semantic search, paper inspection, citation‑graph expansion, and reading specific passages.	Sub‑commands: search-papers, inspect-paper, related-papers, read-paper, search-github
login	Opens the browser for OAuth login or accepts an API key via --api-key.
setup …	Installs skills (CLI primitives, build‑SDK helpers, workflow recipes) into AI coding agents, or configures the MCP (Multi‑Client Provider) server inside editors.

---

Skills & Agent Integration

Firecrawl ships a catalog of skills (firecrawl/skills) that teach AI coding agents how to:

• Use the CLI for live web work.
• Produce ready‑made deliverables (SEO audit, lead list, research brief, etc.).

You can install them in bulk:

firecrawl setup core      # basic scrape/search/crawl primitives
firecrawl setup build    # SDK‑style integration helpers
firecrawl setup workflows# end‑to‑end recipes

Or target a single harness:

firecrawl setup core --agent codex

The setup defaults command can make Firecrawl the default web provider for a given agent, overriding the agent’s native fetch implementation.

---

Authentication & Self‑Hosting

• First run prompts for login (browser or manual API key).
• You can also set FIRECRAWL_API_KEY in the environment.
• For self‑hosted instances, supply --api-url <url> (or FIRECRAWL_API_URL). When a custom URL is used, the CLI skips authentication, allowing local development without a key.

---

Example Workflows

1. One‑off scrape to markdown

   firecrawl https://example.com -o page.md
   

2. Search and automatically scrape the top 5 results as JSON

   firecrawl search "AI agents" --scrape --scrape-formats markdown,links --json -o results.json
   

3. Find biomedical papers about CRISPR and read a specific claim

   firecrawl research search-papers "CRISPR base editing" --limit 10
   firecrawl research read-paper doi:10.1016/j.neunet.2025.108095 \
       --question "What was the sample size?"
   

4. Add Firecrawl skills to the Cursor editor only

   firecrawl init --agent cursor
   firecrawl setup skills --agent cursor
   

---

Who Might Use This?

• Developers building applications that need reliable web‑scraping or search without writing their own crawlers.
• AI agents (Claude Code, Cursor, OpenHands, etc.) that require a plug‑in to fetch live web data.
• Researchers who want quick semantic access to millions of paper abstracts.
• Product teams automating SEO audits, lead generation, or knowledge‑base creation via the ready‑made workflow skills.

---

Bottom Line

The Firecrawl CLI is a thin, terminal‑friendly wrapper around a powerful AI‑backed web‑data service. It handles authentication, offers a rich set of output formats, and can extend AI coding agents with reusable “skills” so that agents can fetch, scrape, and reason over live web content without custom code.