Firecrawl Search

/firecrawl-search

Search the web through Firecrawl and get back ranked results ready for scraping or citation.

Usage

# CLI
firecrawl search "history of trivia game shows" --limit 6

# Portal API (works keyless on the free tier, rate-limited)
curl -X POST "$CF_API/api/skills/firecrawl/search" \
  -H 'Content-Type: application/json' \
  -d '{"query": "history of trivia game shows", "limit": 6}'

What it does

1. Sends the query to Firecrawl /v2/search (SDK when FIRECRAWL_API_KEY is set, keyless REST otherwise).
2. Normalizes results to {title, url, description} rows.
3. Hand promising URLs to /firecrawl-scrape for full markdown, or to tools/research/firecrawl_research.py for citation bundles.

When NOT to use

• You already have the exact URL — go straight to /firecrawl-scrape.
• You need trusted-domain, persisted research for a brand — use /research (pipelines/research/topic.py), which stores wiki_sources rows.

Examples

firecrawl search "best AI video production companies 2026" --limit 10

Implementation

api/routes/skills.py::firecrawl_search and the Firecrawl CLI.