How to integrate Firecrawl MCP with Atomic Agent

Tool to cancel an in-progress agent job by its ID.

Enrich up to ten companies in one call, billed 30 Firecrawl credits per matched company; unmatched records are free.

Use when you have a domain or LinkedIn company URL and need the company behind it.

Look up a company by its Apollo ID; use companies/enrich when you have a domain instead.

Find current hiring activity at a known company.

Find news about known Apollo companies, optionally filtered by category and dates.

Use to build a list from a shape: size, location, industry, technology.

Enrich up to ten identified people in one call, billed 30 Firecrawl credits per matched person; unmatched records are free.

Retrieve a known Apollo person by ID.

Use to resolve one person you have already found.

Tool to scrape multiple URLs in batch with concurrent processing.

Tool to cancel a running batch scrape job using its unique identifier.

Retrieves the current status and results of a batch scrape job using the job ID.

Tool to retrieve error details from a batch scrape job, including failed URLs and URLs blocked by robots.

Use for where analysts stand in aggregate right now on one company.

Use for who is buying whom, at what price, and whether the deal is pending or closed.

Use for how analysts have moved on a stock: upgrades, downgrades, initiations and target changes.

Use to resolve an analyst by name, or to judge how accurate a particular analyst has been, and to get the id you then pass to calendar/ratings as parameters[analyst_id].

Use to enumerate the covering firms or to turn a firm name into the id that calendar/ratings filters on as parameters[firm_id].

Use when you keep a copy of the ratings or M&A calendar and need to know which previously returned events have since been withdrawn, which Schedule C.

Use this for company-issued announcements as distributed over the wires (PRNewswire, Globe Newswire, Business Wire, ACCESS Newswire, Newsfile), the primary-source wording behind a story.

Use this when you keep a copy of Benzinga news and need to know which stories have been withdrawn so you can delete them, which Schedule C.

Use for what was reported about a company or subject, and when.

Use this to answer why a stock moved: WIIMs are the short attributed explanations Benzinga files against a ticker's price action.

Use for positioning: large directional bets before a catalyst.

Initiates a Firecrawl web crawl from a given URL, applying various filtering and content extraction rules, and polls until the job is complete; ensure the URL is accessible and any regex patterns for paths are valid.

Cancels an active or queued web crawl job using its ID; attempting to cancel completed, failed, or previously canceled jobs will not change their state.

Tool to cancel a running crawl job by its ID.

Tool to retrieve the status and results of a Firecrawl crawl job.

Tool to retrieve errors from a Firecrawl crawl job.

Tool to retrieve all active crawl jobs for the authenticated team.

Preview crawl parameters before starting a crawl by generating optimal configuration from natural language instructions.

[NEW v2 API] Initiates a Firecrawl v2 web crawl with enhanced features over v1: natural language prompts for automatic crawler configuration, crawlEntireDomain for sibling/parent page discovery, better depth control with maxDiscoveryDepth, subdomain support, and full webhook configuration.

Tool to get current team credit usage information.

Tool to retrieve historical team credit usage on a monthly basis.

Use the base Data Legion product by default for company enrichment.

Use only when the caller explicitly requests Data Legion premium company enrichment; otherwise use the base action.

Use the base Data Legion product by default for person enrichment when personal contact data is not wanted.

Use the base Data Legion product when the caller explicitly requests personal contact data.

Use only when the caller explicitly requests premium Data Legion person enrichment without personal contact data.

Use only when the caller explicitly requests both premium Data Legion person enrichment and personal contact data.

Extracts structured data from web pages by initiating an extraction job and polling for completion; requires a natural language `prompt` or a JSON `schema` (one must be provided).

Tool to retrieve the status and results of a previously submitted extract job.

Call to discover what is covered, or to resolve names to identifiers when you do not already have a ticker.

Call first when starting from a ticker or stable company or security identifier.

Call for the numbers a business is actually run on - segment revenue, units, subscribers - which never reach the three statements.

Call when a per-share figure has to be computed, or when a dual-class structure means the headline share count is misleading.

Call to answer who reports when.

Call to find what exists for a quarter before asking for any of it - the deck, the release, the report, the transcript.

Call for what was said on the call, attributed and timed.

Call to find out what exists and when it was filed, and to get the document links.

Call when comparing against analyst estimates or consensus, which are quoted on an adjusted basis rather than a GAAP one.

Call when comparing companies, screening, or computing anything that has to mean the same thing across two filers.

Call when fidelity to the filing matters more than comparability: the labels and groupings are the company's own, so two companies cannot be lined up against each other.

Call when comparing companies, screening, or computing anything that has to mean the same thing across two filers.

Call when fidelity to the filing matters more than comparability: the labels and groupings are the company's own, so two companies cannot be lined up against each other.

Call when comparing companies, screening, or computing anything that has to mean the same thing across two filers.

Call when fidelity to the filing matters more than comparability: the labels and groupings are the company's own, so two companies cannot be lined up against each other.

Call to find which companies professional investors are actually writing about, and how heavily, before asking what they said.

Call for the qualitative case on a company as professional investors put it, with the stance and conviction attached.

Call for what one letter actually argued.

Call to understand how a firm invests before weighting what it says about a company.

Call to find the investorId for a firm, or to pick firms by style or sector focus before reading any letters.

Call to see what was published in a quarter.

Call to find a ratio id before requesting a series, and to check `hasDailyData` before asking for a daily one - not every ratio has it.

Call to find the standardized metric id for a concept before requesting statements, rather than guessing at a name.

Call rather than the full list when the template is already known - it is the only place the prose definition and the sub-component breakdown of each metric are published, which is what settles what a line actually includes.

Call to read ownership from the investor's end rather than the company's - what a fund holds and how it is weighted.

Call to resolve an institution's name to the holder id that ownership/holder-holdings needs.

Call for the current standing position of officers and directors, including holdings held through trusts and other vehicles.

Call to see insider activity over time.

Call to see who institutionally owns a company and who moved last quarter.

Call for a daily close series.

Call when the question is about movement within a session - a reaction to an announcement, say.

Call to explain a discontinuity, or to reconcile against a price series from elsewhere that is not split-adjusted.

Call when the question is how a ratio moved between reporting dates rather than what it was at them.

Call for valuation, margin and leverage ratios already computed against the reported statements, rather than recomputing them from financials and risking a different denominator.

Resolve one company from its domain, LinkedIn company URL, or LinkedIn company ID.

Use only for strict exact-match semantics with a LinkedIn company URL.

Find companies by industry, size, location, technologies, or founding year.

Find a person's mobile phone number from a LinkedIn URL, or from their name and employer.

Find a person's personal email from a LinkedIn URL, or from their name and employer.

Identify the person and company behind a work or personal email address.

Find a person's verified work email from a LinkedIn URL, or from their name and employer.

Resolve a person from a LinkedIn URL or ID, or a full name combined with a company domain, LinkedIn URL, or ID.

Use only for strict exact-match semantics with a LinkedIn profile URL.

Tool to get the status and results of an agent job.

Retrieves the status and results of a deep research job by its ID.

Retrieves the current status, progress, and details of a web crawl job, using the job ID obtained when the crawl was initiated.

List standalone Firecrawl Interact browser sessions, optionally by status.

List recent check runs for a Firecrawl monitor.

Initiates an async job to generate an LLMs.

Tool to get the status and results of an LLMs.

Maps a website by discovering URLs from a starting base URL, with options to customize the crawl via search query, subdomain inclusion, sitemap handling, and result limits; search effectiveness is site-dependent.

Resolve a canonical Instagram post/video URL against Oriane's index.

Search Oriane's indexed Instagram posts, captions and engagement.

Pass a canonical Instagram profile URL to look up account details in Oriane's index.

Search Oriane's indexed Instagram profiles, bios and follower counts.

Search Oriane's indexed TikTok posts, captions and engagement.

Pass a canonical TikTok profile URL to look up account details in Oriane's index.

Search Oriane's indexed TikTok profiles, bios and follower counts.

Resolve a canonical TikTok post/video URL against Oriane's index.

Use for the company record behind a sponsor, an entity or a companies/search hit.

Use for a company's podcast advertising footprint in one call, across all its brands.

Use for the feed of a company's actual ad reads, episode by episode, to read or audit them.

Use as a buy-side prospect list: where an advertiser could go next, with the shows it already buys as the reason.

Use for where a company advertises, ranked by volume, with the reads to sample.

Use for who a sponsor shares ad breaks with: the advertisers that buy the same inventory.

Use for exactly who advertised in one episode, with the offer and where in the episode the read sits.

Use for the biggest podcast advertisers overall, in a period, or on one network.

Use for a quick read of who is spending most on podcasts this week and who is climbing.

Use for how a network is monetised and who buys across it.

Use for which of a network's shows carry the advertising.

Use for which networks carry the most advertising, or monetise their catalogue most densely.

Use for who buys a network, and which of them buy it as a bundle: sort by podcast_coverage with a minimum.

Use for who sponsors a show and how heavily it is monetised.

Use as a prospecting list for a show selling its own inventory: who buys shows like mine and has not bought me.

Use for every advertiser on one show with counts and recency, or one company's brands on it.

Use to find who is buying advertising, or to go from a sponsor back to the company paying for it.

Use to tell network buys from per-show buys: a sponsor at 60 to 100 percent coverage of several publishers is buying bundles.

Use to find a sponsor by name or by company and pick up its id.

Use to read the actual ad reads for a sponsor, then podcasts/segment/transcript for the words of one.

Use for where an advertiser buys: the canonical footprint of one sponsor, or of a whole company across its brands.

Use for which advertisers are ramping podcast spend right now, by airdate.

Use for how an advertiser's podcast spend trends by week or month, optionally on one show or network, and its host-read mix.

Use for a network's political profile as a whole: how political its catalogue is and which way it leans.

Use for which publishers carry the most shows of one lean: the flip of a publisher's own bias profile.

Use to list a network's shows by political lean, or only the ones in a bucket.

Use for which networks lean furthest left or right, are most political, or most ideologically diverse.

Use for where a show sits politically and why, with the evidence to quote.

Use for what podcasts are talking about most this week in one domain: the most-mentioned shows, films, games, teams or players, or the most-booked guests, with movement since last edition.

Use to see which charts exist (all, tv-series, movies, video-games, nfl, nba, guests and so on) and who leads each, before fetching one with charts/chart.

Use for a company's competitive set with the reason each one is on it, before comparing their podcast presence or advertising.

Use to get a company's handles, domain, CIK or ticker.

Use to find executives and role holders at a company, or the marketing, brand and partnerships contacts a sponsorship conversation would go to.

Use to see what a company sells, structured by segment and product line, for example to match an ad read's offer to a product.

Use to resolve a company by name, ticker, domain, CIK or Wikidata QID to the slug every company capability takes.

Resolve one entity slug or ID to its knowledge-graph record and linked company or person.

Use for the most-discussed entities overall, in one show, or of one kind, or to fetch several entities by slug at once.

Use when you hold a LinkedIn slug, a handle, a domain, a ticker or a CIK and need the Particle person or company: the reverse of the external-links capabilities.

Use to turn what someone typed into a specific person, company or entity and its slug, before asking what was said about it with podcasts/mentions or filtering episodes by it.

Use to place a topic in the taxonomy and see its neighbours.

Use to find the topic_id that podcasts/search, podcasts/list and the guest capabilities filter on.

Use to see what the type filter on entities/list can take.

Use to get a person's LinkedIn or social handles.

Use when the question is about someone as a podcast guest: how often they appear, where, and since when.

Use for where someone has appeared, episode by episode, to read what they said: take episode.

Use to build a pitch list for a guest: the shows most like the ones that already booked them, with the venues to cite.

Use to find guests by name, by show, by topic or by how often they appear, and pick up the person slug.

Use for the set of shows behind a guest rather than the episodes: which shows book them and how often.

Use for who is doing the podcast circuit right now: a book launch, a product unveil, a news moment, or someone new.

Use to go from a name on an episode to the person: who they are, where they work and worked, and their profile links.

Use for who has been on a show and how often.

Use as a booking pipeline for a show: guests its peers chose that it has not had yet.

Use when a segment needs to be cited as a playable range rather than as text.

Use to browse quotable moments by show, speaker or kind.

Use to quote a clip verbatim with speaker labels, or export it as subtitles with format srt.

Use episode.

Use for the quotable moments of an episode as bounded, playable ranges with a kind and a speaker.

Use for what an episode was about in named things: the companies, people and products it discussed, ranked by salience.

Use for who else covered this: the same story or subject on other shows, optionally within a week of the episode.

Use to see the structure of an episode and pick the section to read: which time ranges are ads to skip, where the interview starts, what each discussion covers.

Use to poll for new episodes of a set of shows or topics as they are transcribed: a resumable, strictly ordered pull.

Use for episode-level filtering without dialogue: every episode a person spoke on, every episode featuring a company, a show's episodes in a date range, episodes in a language.

Use when you hold a Spotify, Apple Podcasts or YouTube episode link, a platform episode id, or an RSS guid and need the Particle episode id to read its transcript.

Use to find who was on an episode and in what role, with the person slug to follow into people/profile or people/guest.

Use to find episodes that discussed something.

Use for a trend rather than a list: how often a person appeared by month, how a show's output changed, how many episodes discussed a term each week.

Use to classify an episode by subject area, or to pick up a topic_id for filtering shows and episodes.

Use to browse shows by attribute rather than by name: a topic, a language, a brand-suitability tier, a length band, a cadence, an ad-free or interview format.

Use when you already hold an Apple collection id, Spotify show id, YouTube channel id or RSS feed URL and need the Particle podcast deterministically.

Use for every line of dialogue about one person or company across podcasts: the read-everything-about-X call.

Use for how mentions of a person or company trend by day, week or month, optionally within one show or network.

Use to go from a show's publisher reference to the network record.

Use to find a network's slug by name before asking about its shows, advertising, bias or suitability.

Use to walk a network's catalogue by audience.

Use to turn a show name into its slug and record.

Use to see what a matched segment is (an interview, a topic discussion, an ad) and its time range.

Use to pull one kind of section across a show: every interview, every ad read, every topic discussion in a date range.

Use to read the words of one segment in full and in order, at the price of a segment rather than an episode.

Use when a search or an episode record gave you a podcast id and the full show record is needed, including the publisher to go up to and the slug every other show capability takes.

Use to get from a show to its episodes and their ids: the latest episode, an episode by title, or the run of episodes in a date range.

Use to get a show's Apple, Spotify or YouTube ids and links, its social handles, or its website.

Use for the exact rates and distributions behind a show's format: how often it has guests, how long episodes run, how often it publishes and on which days.

Use for how much one show talks about a person or company: an episode-level roll-up inside one podcast.

Use to read what listeners wrote about a show.

Use for a show's average rating and count per storefront, plus the combined figure and what recent reviews say in aggregate.

Use for shows like this one: competitive sets, a listener's next show, or the peer set a sponsor or guest pitch cites.

Use to see which subject areas have the most shows.

Use to read a whole episode rather than the excerpt a search returned.

Use to see exactly where and how a company, person or product was discussed within one episode, with the dialogue around each mention.

Use to read what was being said at a timestamp, or the opening of an episode, at a fifth of the full transcript's price.

Use to find the category_slug the chart capabilities take.

Use for today's chart: the US Apple overall top podcasts with no arguments, or a country and category.

Use to find the country codes the chart capabilities take.

Use for how a chart looked on past days, or how one show moved within it.

Use for what debuted, climbed, fell or dropped off a chart since yesterday or over a window.

Use for where a show charts today, everywhere it charts.

Use for a show's chart trajectory over time.

Use for how big a show is on the charts in one call: its best rank and how widely it charts.

Use to check which chart sources exist and how fresh the snapshot is.

Use to pivot on one category for a regulated or family brand: which networks carry the most, or least, alcohol, weapons, adult or hate-speech exposure.

Use for what kind of shows a guest tends to appear on, in brand-safety terms.

Use for how brand-safe a network's catalogue is as a whole and where its exposure sits.

Use to build an inclusion or exclusion list inside one network: its shows by tier, or the ones exposed in a category.

Use for the safest or riskiest networks to buy, or the most broadly placeable.

Use for whether a show is safe to advertise on and in which categories it is exposed, with the evidence.

Use when you have one company and want its firmographics, funding and headcount.

Use to find companies by attribute - industry, headcount, funding stage, location - rather than to look one up you can already name.

Tool to retrieve metrics about the team's scrape queue.

Scrapes a publicly accessible URL, optionally performing pre-scrape browser actions or extracting structured JSON using an LLM, to retrieve content in specified formats.

Performs a web search for a query, scrapes content from the top search results using Firecrawl, and returns details in specified formats.

Search professional profile previews by role, company, keywords and location through Firecrawl Alexandria Apollo.

Search professional profiles by current or past roles, employers, skills, education and location through Firecrawl Alexandria FullEnrich.

Use for marketing channel shares for one complete month.

Use for marketing channel traffic for one complete month.

Use for traffic geography for one complete month.

List domains sending referral traffic to the supplied domain for one complete month, using record-offset pagination.

Use for paid search spend for one complete month.

List domains sending referral traffic to the supplied domain for one complete month, using page-number pagination.

Use for traffic and engagement for one complete month.

List domains that receive referral traffic from the supplied domain, optionally filtered by date, country, device, and granularity.

Tool to start an agent job for agentic web extraction with multi-page navigation and interaction capabilities.

Stop a scrape job's interactive browser session, releasing browser resources and finalizing duration-based billing.

Tool to retrieve the current team's token usage and balance information for Firecrawl's Extract feature.

Tool to retrieve historical team token usage on a monthly basis.

Use for how a market advertises: the share of hooks, personas, offers or formats across a vertical, brand or filter.

Use to list a brand's or a vertical's ads by filter: what an advertiser is running now (advertiser_id + is_active=true), or the winners in a category (order_by=best, min_impressions=4).

Use for everything about one ad: the hook, offer, landing page and whether the page delivers what the ad promises.

Use to find Meta ads matching a free-text idea, relevance-ranked across Upspring's competitor ad corpus.

Use to find ads creatively like one ad, from the same brand and from its competitors.

Use for a brand's ad strategy at a glance: size grades, Facebook and Instagram followers, Reddit sentiment, its creative fingerprint (personas, mechanisms, hooks, media mix, funnel), hook themes and its top ads.

Use when all you have is a company's website: resolves it to one advertiser and returns the same brief as advertisers/brief.

Use to put two brands' ad strategies side by side: where they overlap, where they differ, and what each leaves open.

Use for brands like this one: same category, vertical, geography or size band.

Use when advertisers/brief answers not found or status unavailable for an advertiser id you know exists: asks Upspring to source it, then poll advertisers/brief.

Use to find a brand's advertiser id by name, website or Facebook page, or to list the advertisers in a vertical, category or size tier.

Use for a brand's closest competitors: advertisers going after the same audience with comparable creative strategies.

Use, free, before filtering ads by persona, hook or another dimension: lists the valid ids in one vertical.

Use first, free, to learn the vertical and category ids the ads and advertiser filters accept.