Firecrawl

Live web search, any URL as clean markdown, site maps, and curated research, developer and government indexes.

Provider id: firecrawl
Tools: 9, at $0.005–$0.01 per call
Categories: Web Search, Content Extraction, Research & Papers, Developer, Government & Law
Homepage: www.firecrawl.dev
Provider docs: docs.firecrawl.dev

Tool	Price	What it does
`firecrawl/ask-page`	$0.01	Ask a question about one web page and get a short answer read from that page.
`firecrawl/developer-search`	$0.005	Search GitHub issues, merged pull requests, READMEs and curated docs, with the matching passages.
`firecrawl/gov-search`	$0.005	Search US federal, state and local primary law: statutes, regulations, codes and court opinions.
`firecrawl/map`	$0.005	List the URLs on a website (optionally only those relevant to a few words), fast.
`firecrawl/read-paper`	$0.005	Open one paper: its metadata, or the passages inside it that answer a question.
`firecrawl/related-papers`	$0.005	Papers similar to, citing, or cited by a given paper.
`firecrawl/research-papers`	$0.005	Semantic search over research papers (arXiv and more): ids, titles, abstracts and scores.
`firecrawl/scrape`	$0.005	Turn any URL (including JavaScript-heavy pages and PDFs) into clean markdown.
`firecrawl/search`	$0.01	Search the live web or news and optionally get every result's page as markdown in the same call.

Inputs are JSON bodies, and unknown fields are rejected with 422 invalid_input (not charged). POST /v1/inspect with a tool's id returns the same input schema, plus the output schema.

Firecrawl Ask a Page

firecrawl/ask-page · $0.01 per call (premium) · AI Answers, Content Extraction

Ask a question about one web page and get a short answer read from that page.

Reads the page in a browser and answers the question from its content only (it does not search the web). Good for 'what does this pricing page say about X'. For questions that need several sources use akashi/answer.

Input

Field	Type	Required	Description
`url`	string (uri)	yes	The page that holds the answer. 1–2083 characters.
`question`	string	yes	What you want to know. 3–500 characters.

Example

Input

{
  "url": "https://www.firecrawl.dev/pricing",
  "question": "How many credits does the free plan include?"
}

Run it with the CLI

node akashi.mjs run firecrawl/ask-page --input '{"url":"https://www.firecrawl.dev/pricing","question":"How many credits does the free plan include?"}'

Firecrawl Developer Search

firecrawl/developer-search · $0.005 per call (standard) · Developer

Search GitHub issues, merged pull requests, READMEs and curated docs, with the matching passages.

One index over public repositories (issues, merged PRs, READMEs) and documentation sites, each hit with the passage that matched, in markdown. Ideal for 'has anyone hit this error' and 'how do I configure X'. For package metadata use npm/package or pypi/package.

Input

Field	Type	Required	Description
`query`	string	yes	A developer question, e.g. 'configure retries in httpx'. 2–500 characters.
`types`	array of string	no	Limit to docs, issues, merged PRs or READMEs. Each item: One of `doc`, `issue`, `pull_request`, `readme`.
`language`	string	no	Repository language, e.g. 'Rust'. At most 40 characters.
`k`	integer	no	From 1 to 20. Default `8`.

Example

Input

{
  "query": "how to configure retries in httpx",
  "k": 5
}

Run it with the CLI

node akashi.mjs run firecrawl/developer-search --input '{"query":"how to configure retries in httpx","k":5}'

Firecrawl Government Search

firecrawl/gov-search · $0.005 per call (standard) · Government & Law

Search US federal, state and local primary law: statutes, regulations, codes and court opinions.

Searches an index of government sources only (fda.gov, ecfr.gov, state codes, court opinions…), so results are primary material rather than law-firm blogs. US only. Not legal advice; read the source.

Related: firecrawl/scrape.

Input

Field	Type	Required	Description
`query`	string	yes	A legal or regulatory question. 2–500 characters.
`k`	integer	no	From 1 to 20. Default `8`.

Example

Input

{
  "query": "food labeling requirements for allergens",
  "k": 5
}

Run it with the CLI

node akashi.mjs run firecrawl/gov-search --input '{"query":"food labeling requirements for allergens","k":5}'

Firecrawl Site Map

firecrawl/map · $0.005 per call (standard) · Content Extraction

List the URLs on a website (optionally only those relevant to a few words), fast.

Discovers a site's URLs from its sitemap and links without scraping them, with titles where known. Use it to find the right page, then read that page with firecrawl/scrape.

Related: firecrawl/scrape.

Input

Field	Type	Required	Description
`url`	string (uri)	yes	Any page on the site to map. 1–2083 characters.
`search`	string	no	Only URLs relevant to these words. At most 200 characters.
`limit`	integer	no	From 1 to 200. Default `50`.

Example

Input

{
  "url": "https://docs.firecrawl.dev",
  "search": "rate limits",
  "limit": 10
}

Run it with the CLI

node akashi.mjs run firecrawl/map --input '{"url":"https://docs.firecrawl.dev","search":"rate limits","limit":10}'

Firecrawl Read Paper

firecrawl/read-paper · $0.005 per call (standard) · Research & Papers

Open one paper: its metadata, or the passages inside it that answer a question.

With only a paper_id it returns the paper's metadata. Add a question to get the in-body passages (text and tables) that answer it, scored by similarity — quote those rather than the abstract.

Input

Field	Type	Required	Description
`paper_id`	string	yes	paper_id or primary id (e.g. arxiv:2510.21538). 1–200 characters.
`question`	string	no	Return the passages that answer this. At most 500 characters.
`passages`	integer	no	From 1 to 12. Default `4`.

Example

Input

{
  "paper_id": "arxiv:2510.21538",
  "question": "How is hallucination detected?"
}

Run it with the CLI

node akashi.mjs run firecrawl/read-paper --input '{"paper_id":"arxiv:2510.21538","question":"How is hallucination detected?"}'

firecrawl/related-papers · $0.005 per call (standard) · Research & Papers

Papers similar to, citing, or cited by a given paper.

Expands from one paper_id to its neighbourhood: 'similar' (by content), 'citers' (newer work that cites it) or 'references' (what it cites). Use after firecrawl/research-papers.

Related: firecrawl/read-paper.

Field	Type	Required	Description
`paper_id`	string	yes	1–200 characters.
`mode`	string	no	similar work, papers citing it, or its references. One of `similar`, `citers`, `references`. Default `"similar"`.
`k`	integer	no	From 1 to 25. Default `8`.

Input

{
  "paper_id": "arxiv:2510.21538",
  "mode": "similar",
  "k": 5
}

Run it with the CLI

node akashi.mjs run firecrawl/related-papers --input '{"paper_id":"arxiv:2510.21538","mode":"similar","k":5}'

Firecrawl Research Papers

firecrawl/research-papers · $0.005 per call (standard) · Research & Papers

Semantic search over research papers (arXiv and more): ids, titles, abstracts and scores.

Ranks papers by meaning, not keywords, so 'why RAG models hallucinate' finds the right work. Each result has a paper_id usable with firecrawl/read-paper (passages inside the paper) and firecrawl/related-papers (similar, citing or cited work). For DOI metadata use crossref/works.

Input

Field	Type	Required	Description
`query`	string	yes	Topic, method, benchmark or author. 2–500 characters.
`k`	integer	no	Papers to return. From 1 to 25. Default `8`.
`authors`	string	no	Comma-separated author names to filter by. At most 200 characters.
`published_from`	string	no	Earliest date, YYYY-MM-DD.
`published_to`	string	no	Latest date, YYYY-MM-DD.

Example

Input

{
  "query": "retrieval augmented generation hallucination detection",
  "k": 5
}

Run it with the CLI

node akashi.mjs run firecrawl/research-papers --input '{"query":"retrieval augmented generation hallucination detection","k":5}'

Firecrawl Scrape

firecrawl/scrape · $0.005 per call (standard) · Content Extraction

Turn any URL (including JavaScript-heavy pages and PDFs) into clean markdown.

Renders the page in a real browser and returns its main content as markdown with the title, description, language and the HTTP status the site answered. Works on single-page apps and PDFs. One page per call: to find pages first use firecrawl/map (one site) or firecrawl/search (the web). Very long pages are trimmed to Akashi's size cap.

Input

Field	Type	Required	Description
`url`	string (uri)	yes	The page to read. 1–2083 characters.
`main_content_only`	boolean	no	Drop navigation, footers and ads. Default `true`.

Example

Input

{
  "url": "https://en.wikipedia.org/wiki/Solid-state_battery"
}

Run it with the CLI

node akashi.mjs run firecrawl/scrape --input '{"url":"https://en.wikipedia.org/wiki/Solid-state_battery"}'

Firecrawl Web Search

firecrawl/search · $0.01 per call (premium) · Web Search, News

Search the live web or news and optionally get every result's page as markdown in the same call.

Live web or news search (not a cached index): titles, URLs and snippets, newest-first when a time range is set. Set with_content to receive each result's main content as markdown, which saves a separate scrape per page. For a single known URL use firecrawl/scrape; for a cited answer to a question use akashi/answer; for cheaper plain Google results use serper/search.

Input

Field	Type	Required	Description
`query`	string	yes	What to search for, in plain words. 1–500 characters.
`limit`	integer	no	Results to return. From 1 to 10. Default `5`.
`source`	string	no	web pages or news articles. One of `web`, `news`. Default `"web"`.
`time_range`	string	no	Only results from the last day/week/month/year. One of `day`, `week`, `month`, `year`.
`country`	string	no	ISO country code, e.g. 'us'. 2–2 characters.
`with_content`	boolean	no	Also return each result's page as markdown (costs more). Default `false`.

Example

Input

{
  "query": "solid-state battery startups",
  "limit": 5
}

Run it with the CLI

node akashi.mjs run firecrawl/search --input '{"query":"solid-state battery startups","limit":5}'

On this page