Firecrawl compatibility

webx implements Firecrawl-shaped routes so existing integrations keep working when you repoint the base URL.

Compatible routes

POST /v1/scrape · POST /v1/crawl · POST /v1/search · POST /v1/map · POST /v1/extract · POST /v2/scrape · POST /v2/crawl · POST /v2/search · POST /v2/map · POST /v2/batch/scrape · POST /v2/extract · POST /v2/agent

Request shapes accepted: formats (including object forms), includeTags, waitFor, actions, maxAge, webhook, limit, and the rest of the documented fields — unknown fields are ignored, not rejected.

Migration

# before — Firecrawl cloud
from firecrawl import FirecrawlApp
app = FirecrawlApp(api_key="fc-…")

# after — self-hosted webx (same SDK!)
app = FirecrawlApp(api_url="http://localhost:8080", api_key="webx_…")

// JavaScript — same shape
import Firecrawl from "@mendable/firecrawl-js";
const app = new Firecrawl({ apiUrl: "http://localhost:8080", apiKey: "webx_…" });

That’s the entire diff — one line.

What translates

Firecrawl feature	webx behavior
formats: ["markdown","html","links","screenshot"]	Same names; webx adds chunks, a11y, agent_ready, transcript, redact_pii, pdf, mhtml
actions	webx --actions grammar — click/fill/scroll/eval/act:<english>
waitFor	--wait-for <selector> equivalent
maxAge	max_age — serves fresh index copies inside the window
webhook on crawl	webhook_url + optional webhook_secret HMAC signing
changeTracking	change_tracking — diff vs indexed version
Async /crawl job polling	GET /crawl/{id} + SSE at GET /crawl/{id}/events

What doesn’t (be honest)

• Firecrawl’s managed scale — their browser fleet and proxy infra. webx renders with your own engine (or WEBX_CDP_URL → a browser fleet you control).
• Credit semantics — webx meters per-key monthly_units on your deployment, not their credit pool.
• Brand-new Firecrawl fields may lag — the compat layer covers the documented surface, not unreleased features.

Full comparison: vs Firecrawl.