Crw Migrate

crw-migrate — Coming from Firecrawl?

Switch in one line. The v2 API is a drop-in for the official firecrawl-py v4 SDK and firecrawl-js — swap the base URL and keep your code.

When to use

• You have existing Firecrawl SDK code or REST calls you want to repoint.
• You're replacing a firecrawl-mcp-server entry in your MCP config.
• You want to know exactly which Firecrawl endpoints are covered vs. which need adaptation.

The one-line swap

firecrawl-py v4 (Python SDK)

Managed fastCRW:

from firecrawl import FirecrawlApp

app = FirecrawlApp(
    api_url="https://api.fastcrw.com",
    api_key="crw_live_..."
)

Self-hosted fastCRW (default port 3000, no auth):

app = FirecrawlApp(
    api_url="http://localhost:3000",
    api_key="any"          # self-host ignores the key when auth is not configured
)

firecrawl-js (TypeScript/Node SDK)

import FirecrawlApp from "@mendable/firecrawl-js";

const app = new FirecrawlApp({
  apiUrl: "https://api.fastcrw.com",   // or "http://localhost:3000"
  apiKey: "crw_live_...",
});

REST / curl

Replace https://api.firecrawl.dev with https://api.fastcrw.com (or your self-hosted http://localhost:3000). Auth header stays the same: Authorization: Bearer <key>.

# Before
curl -X POST https://api.firecrawl.dev/v1/scrape \
  -H "Authorization: Bearer crw_live_..." \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com","formats":["markdown"]}'

# After — change exactly two things: host + key
curl -X POST https://api.fastcrw.com/v1/scrape \
  -H "Authorization: Bearer crw_live_..." \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com","formats":["markdown"]}'

Compatibility matrix

Drop-in endpoints (no code change needed)

Gaps — what needs adaptation

jsonSchema alias

Firecrawl's /v1/extract uses extract.schema. fastCRW's /v1/scrape accepts both the jsonSchema top-level field and the extract.schema alias for closer Firecrawl parity:

{
  "url": "https://example.com",
  "formats": ["json"],
  "jsonSchema": {
    "type": "object",
    "properties": { "title": { "type": "string" } }
  }
}

Requires [extraction.llm] configured in config.toml (or CRW_EXTRACTION__LLM__API_KEY env var). See crw-self-host.

Switching your MCP config

Firecrawl MCP uses firecrawl-mcp-server; fastCRW's MCP server is crw-mcp. Tool names change from firecrawl_* to crw_*:

Claude Code — replace in ~/.claude/claude_desktop_config.json (or settings)

{
  "mcpServers": {
    "crw": {
      "command": "npx",
      "args": ["crw-mcp"],
      "env": {
        "CRW_API_URL": "https://api.fastcrw.com",
        "CRW_API_KEY": "crw_live_..."
      }
    }
  }
}

For self-hosted (no auth, no env needed):

{
  "mcpServers": {
    "crw": {
      "command": "npx",
      "args": ["crw-mcp"]
    }
  }
}

Embedded mode (npx crw-mcp with no CRW_API_URL) runs the engine in-process — zero server to stand up, ~14 MB RAM.

Verify the swap — checklist

Run these after pointing at fastCRW. Each should return "success": true:

# 1. Health check (no auth)
curl http://localhost:3000/health
# → {"status":"ok",...}

# 2. Basic scrape
curl -X POST "$CRW_API_URL/v1/scrape" \
  -H "Authorization: Bearer $CRW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com","formats":["markdown"]}' | jq .success
# → true

# 3. Map (URL discovery)
curl -X POST "$CRW_API_URL/v1/map" \
  -H "Authorization: Bearer $CRW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com"}' | jq '.data | length'
# → N (should be > 0)

# 4. Search (requires a search backend — managed always works; self-host needs sidecar)
curl -X POST "$CRW_API_URL/v1/search" \
  -H "Authorization: Bearer $CRW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"fastCRW scraper","limit":3}' | jq .success
# → true

# 5. Structured extraction (requires [extraction.llm] configured)
curl -X POST "$CRW_API_URL/v1/scrape" \
  -H "Authorization: Bearer $CRW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url":"https://example.com",
    "formats":["json"],
    "jsonSchema":{"type":"object","properties":{"title":{"type":"string"}}}
  }' | jq '.data.json'

Compare the data.markdown / data.metadata shape from your existing Firecrawl responses — the field names on the overlap surface (title, description, sourceURL, statusCode) match. A few metadata sub-fields diverge; inspect with jq .data.metadata if your code reads specific keys.

See also

• crw-self-host — stand up your own crw server + search backend
• crw-best-practices — SDK patterns, error handling, batching
• crw — hub skill, full verb ladder