Firecrawl clone

Scrape, crawl, map, and search for web-ingestion agents. Error responses use recorded fixture shapes — no live API calls.

Clone ID: firecrawl · Tier: fixture · Binary: clones-fixture

Start here

Question	Answer
Best for	Agents that scrape pages, start crawls, or run web search via Firecrawl.
Connect with	Route mode for `api.firecrawl.dev`, REST at `{base}/firecrawl/api/...`, hosted `https://{sessionId}.clones.molar.it/firecrawl/api`, or MCP `molar_clone_spawn`
Known limits	Recorded responses only; unimplemented paths return `501`; read-only by default unless promoted stateful
Seeds	`empty` — apply on spawn or via `molar_clone_seed`

MCP tools

Fixture clones use shared control-plane tools — no vendor-specific MCP surface.

Tool	Description
`molar_clone_spawn`	Spawn fixture session (`kind: "firecrawl"`, optional `seed`)
`molar_clone_seed`	Apply a named seed to an existing session
`molar_clone_route`	Map `api.firecrawl.dev` to the active session
`molar_clone_stop`	Tear down the session

See MCP server for world_snapshot, world_restore, and cross-clone tools.

REST API coverage

The clone handles these recorded endpoints (local base http://localhost:8000/firecrawl/api):

GET /v1/crawl/crawl_7a8b9c0d1e2f
POST /v1/crawl
POST /v1/map
POST /v1/scrape
POST /v1/search

If your agent calls a path not in this list, the clone returns 501 unimplemented_in_clone and logs the request. Submit missing coverage via the dashboard Developers → 501 request form.

Seeds

Seed	When to use
`empty`	Starting state for minimal / blank workspace scenarios

molar clones spawn firecrawl --seed empty

Connect

Mode	URL pattern
Local REST	`http://localhost:8000/firecrawl/api{path}`
Hosted REST	`https://{sessionId}.clones.molar.it/firecrawl/api{path}`
Route mode	Point `api.firecrawl.dev` at the clone via `molar_clone_route`
Dashboard	Clones explorer — filter by `firecrawl`

Set MOLAR_CLONE_FIRECRAWL_URL for BYOC fleets.

See Fixture clones overview and API & SDK.

Discord GitHub