
Firecrawl Local MCP Server: Setup for Claude, Cursor and Other Apps
Lets your assistant scrape pages, search the web, crawl whole sites and watch pages for changes, with the server running on your computer.
This is Firecrawl's own server, run from your computer with npx or Docker. It sends your scrape, search and crawl requests to Firecrawl's cloud with your API key, or to a Firecrawl you host yourself. Firecrawl also runs a hosted version with a keyless free tier, which is quicker to set up. Pick the local one if your app only starts local servers or you point it at your own Firecrawl.
Good for
- Developers who turn websites into clean text for AI
- Researchers who search and crawl many sites in one request
- Teams that host their own Firecrawl and want the assistant to use it
Works in
12 should work
Tap an app to see how to set it up and any known problems.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Client does not support this transport.
Client does not support this transport.
Install firecrawl-mcp-server
Claude Code
- 1
Copy this prompt.
Add the firecrawl-mcp-server server to Claude Code. Run this command: claude mcp add firecrawl-mcp -e FIRECRAWL_API_KEY=fc-YOUR_API_KEY -- npx -y firecrawl-mcp Ask me for FIRECRAWL_API_KEY and only put the values in that config, nowhere else. Then check that it loaded and tell me if I need to sign in or restart.
- 2
Paste it into Claude Code. It shows the change and asks before running it.
Requires Node.js.
The README also offers a Docker image and a local HTTP mode. Set FIRECRAWL_API_URL to use your own Firecrawl.
From the project README. Source.
Tools
Gets the content of one web page
Scrapes a single page and returns JSON or Markdown.
Try: “Scrape https://example.com/pricing and summarize the plans.”
Network ·
firecrawl_scrapeSearches the web
Runs a web search and returns results.
Try: “Search the web for recent news about MCP servers.”
Network ·
firecrawl_searchLists the URLs on a website
Discovers the pages of a site and returns their URLs.
Try: “List all the pages on docs.example.com.”
Network ·
firecrawl_mapReads many pages of a site
Crawls a site with limits and returns the final status and data after it polls the job.
Try: “Crawl the blog section and collect the post titles.”
Network ·
firecrawl_crawlChecks on a crawl
Returns the status of a crawl job.
Try: “Is my crawl done?”
Network ·
firecrawl_check_crawl_statusTurns a file into Markdown or JSON
Parses PDFs, Word files, spreadsheets and HTML files into Markdown or structured JSON. Local runs can read the file directly.
Try: “Parse this PDF into Markdown.”
Files ·
firecrawl_parseResearches across many sites
Runs an agent that gathers structured data from many or unknown sites.
Try: “Find the pricing of the top five project tools as a table.”
Network ·
firecrawl_agentChecks on an agent run
Returns the status and results of an agent run.
Try: “How is my research run going?”
Network ·
firecrawl_agent_statusClicks and types on a page
Interacts with a URL or a scraped page and returns the result.
Try: “Open this page, click "Load more" and read the new items.”
Write ·
firecrawl_interactEnds a page session
Stops an interact session.
Try: “Close the browser session.”
Write ·
firecrawl_interact_stopSearches developer sources
Answers programming questions with passages from developer sources.
Try: “How do I paginate with the Stripe API?”
Network ·
firecrawl_developer_searchSearches US law and regulations
Searches US primary law and regulatory material.
Try: “What does the law say about data breach notice?”
Network ·
firecrawl_gov_searchSearches research papers
Searches for papers on a topic.
Try: “Find recent papers on retrieval augmented generation.”
Network ·
firecrawl_research_search_papersReads a research paper
Reads the content of a paper.
Try: “Read this paper and summarize the method.”
Network ·
firecrawl_research_read_paperFinds related papers
Finds papers related to one you name.
Try: “Find papers related to this one.”
Network ·
firecrawl_research_related_papersLooks inside a paper
Inspects a paper and its repository matches.
Try: “Does this paper have code on GitHub?”
Network ·
firecrawl_research_inspect_paperBrowses the Alexandria tool catalog
Browses Firecrawl's Alexandria catalog of providers and tool contracts, or looks up a URL.
Try: “What tools can read this site?”
Network ·
firecrawl_find_toolsStarts watching a page for changes
Creates a monitor that checks a page on a schedule and records diffs.
Try: “Watch this pricing page and tell me when it changes.”
Write ·
firecrawl_monitor_createLists your page monitors
Lists your monitors.
Try: “What pages am I monitoring?”
Network ·
firecrawl_monitor_listShows one monitor
Gets one monitor.
Try: “Show the pricing page monitor.”
Network ·
firecrawl_monitor_getChanges a monitor
Updates fields such as the goal, webhook and notification.
Try: “Send alerts for that monitor to my webhook.”
Write ·
firecrawl_monitor_updateChecks a page now
Triggers a monitor check right away.
Try: “Run the pricing monitor now.”
Write ·
firecrawl_monitor_runLists past checks
Lists checks for a monitor, optionally by status.
Try: “List the last checks for that monitor.”
Network ·
firecrawl_monitor_checksShows what changed in one check
Gets page-level results for a check, including the diff.
Try: “What changed in the last check?”
Network ·
firecrawl_monitor_checkDeletes a monitor
Deletes a monitor. The README says to call it only when you mean to remove it.
Try: “Delete the pricing page monitor.”
Delete ·
firecrawl_monitor_deleteShows your credit use
Reports how many Firecrawl credits you have used.
Try: “How many Firecrawl credits have I used?”
Network ·
firecrawl_credit_usageRates a search result
Sends feedback about a search to Firecrawl.
Try: “That search missed the answer, send feedback.”
Write ·
firecrawl_search_feedbackSends general feedback
Sends general feedback to Firecrawl.
Try: “Send feedback that the scrape was incomplete.”
Write ·
firecrawl_feedback
Read from the source code
Permissions
high risk · 3 things to watch
High risk
It reads web pages and uses your Firecrawl credits, and it can create, change and delete page monitors in your account. Set a credit limit in Firecrawl and keep crawls small.
What it can do
- Read your data
- Create and change things
- Delete things
- Read-only mode
Watch out for
It can delete the page monitors in your account
The monitor tools create, update and delete recurring checks. Only ask for a delete when you mean it.
Crawls and agent runs can use up your credits
Crawl and agent tools can fetch many pages in one request. Set limits in the request and watch your usage.
Web pages can carry hidden instructions
Text scraped from a site can try to steer the assistant. Do not give the assistant other tools that change things in the same chat when you scrape pages you do not trust.
Security advisories
No published security advisories as of Nov 1, 2026.
Technical details
| Talks to | Firecrawl API (or your own Firecrawl address). Websites you ask it to open |
|---|---|
| Scopes |
|
Questions
How is this different from the hosted Firecrawl server?
This one runs on your computer and needs your API key. The hosted server is run by Firecrawl and has a keyless free tier for scrape, search and parse. Both reach the same Firecrawl service, and only the local one can point at a Firecrawl you host yourself.
Is it official?
Yes. It comes from Firecrawl's own GitHub organization.
Do I need an API key?
Yes for the cloud service. Get one from the Firecrawl dashboard. If you point it at your own Firecrawl address, the key is optional.
Does it run on my computer?
The server does. Your requests still go to Firecrawl's cloud, or to the Firecrawl address you set.
Can I use it with Docker?
Yes. The README gives a Docker image and also a local HTTP mode. Use the install box for the npx setup.
Can I run it read-only?
No read-only switch is documented.
Information
updated Nov 2, 2026
- Publisher
- Firecrawl (official)
- Package
firecrawl-mcp- Registry
- npm
- Transport
- stdio
- Auth
- API key
- License
- MIT
- Last commit
- Oct 29, 2026
- Open issues
- 101
- Listed in
- Docker MCP Catalog
Other Firecrawl MCP servers
Sources
Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.
Show all 1Hide
- Firecrawl MCP server README · github.com