Connector · Search and research
A Firecrawl teammate in Slack, running on Firecrawl's MCP server.
Chickpea gives a channel a named AI teammate that holds its own Firecrawl connection. It calls Firecrawl's MCP server at mcp.firecrawl.dev with an API key you create in the dashboard, so there is no client to configure. Ask it in Slack to read a page as markdown, list every URL on a site, or search for something. The answer lands in the thread.
What people ask it.
The teammate is @growth, in #growth. Pick a scenario to watch it play.
Read a page as markdown Read only
@growth read northwind.co/pricing and tell me what changed since the version in this thread
@growth summarise this changelog page and say which items affect our API
One URL in, clean text back. Firecrawl documents a
markdownoutput format. It also documents a main-content filter that drops headers, navigation, and footers before the markdown is generated, so the teammate reads the page rather than the furniture. A read needs no confirmation.Map a site, then crawl the part you want Asks first
@growth map northwind.co and show me the docs pages that mention migrations
@growth crawl their whole help centre and pull every claim about uptime
Mapping discovers a site's URLs and is cheap. A crawl is not. Firecrawl charges one credit a page for a basic scrape, crawl, or map, so a large site is a real bill. Ask for a whole site and the teammate comes back with the size it found and waits for a yes before it spends anything.
Search the web Read only
@growth search for what people are saying about northwind pricing this month
@growth find the three best write-ups of the new EU accessibility rules and give me the links
Firecrawl's search tool covers web, news, image, and developer sources, and search bills two credits per ten results. The teammate posts the links it used, so anyone in the channel can check where an answer came from instead of trusting the summary.
Pull structured fields out of a page Read only
@growth get the plan names, seat prices, and trial lengths from northwind.co/pricing as a table
@growth from each of these six job posts, pull the title, the location, and the salary band
Firecrawl takes a JSON schema on a scrape and returns fields that match it. The answer arrives as a table in the thread rather than as prose you have to re-read. Same one credit a page. Nothing is written anywhere until you ask for that separately.
On a schedule Applies at once
@growth every Monday at 8am Pacific, read northwind.co/pricing and post here if anything changed
@growth every weekday at 7am Eastern, search for mentions of us from the last day
A schedule saves on the spot. It needs an explicit time zone, runs under the Firecrawl connection of the person who asked, and is paused rather than run if that connection goes away. Keep the page count small, because every run spends credits.
Connect it
Five steps, from Admin or from Slack.
The credential is asked for once, on a page Chickpea serves, and never in Slack. The link the teammate posts is good for one action, for one teammate, for 24 hours.
- In Firecrawl, open the dashboard's API keys page and create a key. It starts
fc-. Firecrawl's own advice is to keep a key in a secret store and never in the MCP URL, which is what Chickpea does with it. - In Chickpea Admin, open the teammate, then Connections, find Firecrawl, and choose Connect. Or ask in Slack,
@growth connect Firecrawl, and follow the link it posts. - Choose who uses it. Team is one shared key that everyone who can use the teammate acts through, and one credit balance for the channel. Personal means each member pastes their own key and spends their own credits.
- Paste the
fc-…key into the write-only Credential field. It is stored once on your deployment and never returned to the browser. Chickpea then connects tomcp.firecrawl.dev/v2/mcpand records the tools the server answered with. - Ask something small.
@growth read example.com and tell me what it sellsis a good first question, because it is one page and one credit.
A connection belongs to one teammate. Giving a second teammate Firecrawl means a second key, so the support teammate spends its own credits and never inherits the growth teammate's balance.
There is no built-in Chickpea skill for Firecrawl, and none is needed. The tools arrive from Firecrawl's MCP server: Chickpea connects before it saves anything, records what the server answered with, and exposes only those tools. Firecrawl documents firecrawl_scrape for one URL, firecrawl_map to discover a site's URLs, firecrawl_search for web, news, image, and developer sources, and firecrawl_crawl to traverse a site or a section, alongside others it marks as subject to your plan. Firecrawl documents the server, its URL, and its tools.
Beside Firecrawl
What the same teammate usually holds too.
Each is its own connection, granted to the teammate alone. All 37 connectors.
ExaExa finds the pages worth reading, Firecrawl reads them properly
AhrefsWhat a competitor ranks for, next to what their page actually says
Google Search ConsoleYour own search performance, checked against the page that earns it
NotionThe research lands in a Notion page instead of scrolling out of the channel
Questions people ask
Does Firecrawl have MCP?
Yes. Firecrawl runs a hosted MCP server at mcp.firecrawl.dev/v2/mcp over streamable HTTP, authorized with your API key as a bearer token, and documents setup for editors and chat clients. Chickpea connects to that same server, so a teammate in Slack uses it without any of those clients installed.
How do I add Firecrawl MCP to Claude Code or Cursor?
Follow Firecrawl's own docs, since each client keeps its own config file and its own way of holding the key. For Slack there is no config file. Open the teammate in Chickpea Admin, choose Firecrawl, paste the key once, and ask. Or ask "@growth connect Firecrawl" in the channel and follow the link it posts.
Which MCP is best for web scraping?
That depends on what you are doing, and the honest answer is that this page cannot rank them. Firecrawl is the one Chickpea ships a preset for. It has a documented hosted server, tools for scraping, mapping, searching, and crawling, and credits you can watch. What Chickpea adds is the place you use it from: a Slack channel, where the page the teammate read is in the thread for everyone.
Is Firecrawl MCP free?
Firecrawl's pricing page lists a free tier of 1,000 credits a month, refreshed monthly, with no credit card required. Paid plans start at 5,000 credits a month. Chickpea itself adds nothing to that. It is Apache 2.0, runs on your own Cloudflare account, and has no seats and no metering.
What does a crawl cost?
Firecrawl bills in credits, shared across its endpoints. Its pricing page says one credit is one page on a basic scrape, crawl, or map. Search costs two credits per ten results, and advanced formats cost more. So a 400-page site is roughly a 400-credit crawl. That is why the teammate maps first, says the number it found, and waits for a yes. Check firecrawl.dev/pricing for the current rates.
Does it use Firecrawl's MCP server or the API?
The MCP server. Firecrawl is a vendor MCP lane: the deployment connects to mcp.firecrawl.dev/v2/mcp over streamable HTTP and sends your key as a bearer token. Chickpea writes no Firecrawl endpoints of its own, and the connection cannot reach any other address.
Can a page it reads tell the teammate what to do?
No. A retrieved page is task input, never permission, and the rule is enforced rather than left to the model. During a Slack turn Chickpea refuses any MCP tool whose name does not start with an unambiguous read verb unless your own message names that action. Every Firecrawl tool name begins with firecrawl_, so every call has to trace back to something a person asked for in the channel. Text on a crawled page cannot start a second crawl, and it cannot reach another connected account either.
Where does the API key live?
On your deployment, in a write-only Credential field outside the teammate record. It never enters the model's context, the tool arguments the model writes, or Slack. The model picks a connection by id and the key goes in at the last step on the way out. Disconnect the account and the stored secret is removed, after any schedule that depended on it is paused.
Can it crawl behind a login?
Only as far as Firecrawl itself can. Firecrawl's scrape options include a headers field, documented as headers to send with the request, usable for cookies and a user agent, so an authenticated fetch is a request that carries the right header. Chickpea holds one credential for this connector, the Firecrawl key, and none for the sites you point it at. Putting a session cookie for another service into a Slack message is a bad trade, so treat this connector as one for the public web.
Firecrawl or Exa?
Both are connectors, and a teammate can hold both. Firecrawl calls itself the web data API to search, scrape, and interact with the web at scale. Reach for it when you know the URL and want the page as clean markdown or as fields. Exa calls itself the world's data for AI agents and searches an index of the public web. Reach for it when you do not know the URL yet. The usual pairing is Exa to find and Firecrawl to read.
Try it
Give a channel a Firecrawl teammate.
One deploy into your own Cloudflare account, connect Slack, paste one token, ask. Apache 2.0, no seats, no metering.
Or run it on Node. The setup guide covers both.
Last checked
Facts on this page were checked against the product and Firecrawl's own pages on September 17, 2026. This page is also available as markdown at /connectors/firecrawl.md.