Firecrawl
Your own site opened and rendered the way a browser would, page by page — batched rather than one at a time, with a bot-walled page retried once through a rotating proxy — and captured as images while we are there, cookie banner clicked out of frame first.
The facts a reader needs
| What it unlocks | The site capture — and with it logos, colors, typography and screenshots. |
| When it is called | Every paid pull, once, at the start (the website leg gates the four that read its pages). |
| Cost basis | $0.01–0.02 per pull on the standard settings (~10–20 pages, 1 credit = 1 page, plus 8 rendered screenshots, no surcharge for the cookie-consent actions or the anti-bot proxy retry); the deeper settings add ≈$0.02–0.11 for the structured-facts/crawl-fallback/PDF reads |
| Environment | FIRECRAWL_API_KEY |
| Bring your own key | Yes. |
Logos, colors and typography carry no Firecrawl charge of their own — they are read out of pages the website leg already paid for. On the deeper settings, the same account also buys a scoped fallback crawl, a same-account change check, a wider structured-facts read, and a linked PDF read.
Endpoints we call
| Endpoint | Why we call it |
|---|---|
POST /v2/map | Sitemap-aware URL discovery. 1 credit per call. |
POST /v2/batch/scrape | Multiple pages as markdown in one job, polled to completion; raw HTML, a full-page screenshot, cookie-consent actions and an inline structured-facts read on the key URLs. 1 credit per page (+4 credits per key page for structured facts, no surcharge for actions). |
POST /v2/scrape (proxy: "auto") | One retry for a page a bot wall blocked. Verified live 2026-08-19: no credit surcharge on this account. Best-effort proxy/fingerprint rotation, not a challenge-solver. |
POST /v2/crawl (deeper settings) | A small, scoped fallback crawl, only when the sitemap+nav pass returns fewer than 5 links. 1 credit per page crawled. |
POST /v2/scrape (changeTracking, deeper settings) | Git-diff mode only — whether a page changed since Firecrawl's own last snapshot of it. No credit surcharge, measured live 2026-08-19. |
POST /v2/scrape (parsers: pdf, deeper settings) | Reads a linked PDF price list or media kit when discovery finds one. 1 credit per PDF page. |
Checks it powers
ScrapeCreators
Public social profiles and their recent posts across seven platforms (2026-08-19 added Threads), plus Instagram's tagged-posts UGC strip, the ad-transparency archives for TikTok, Google and LinkedIn, and — since the 2026-08-19/20 evidence-quality round — the brand's own YouTube-video transcripts, organic hashtag/keyword mentions, Reddit-community sizing, and LinkedIn follower honesty (all tier >= 2), plus TikTok audience demographics, OCR'd Google ad copy, and permanent Instagram media (all tier >= 3, flag-gated, never on a default pull).
Context.dev
An independent read of the brand's rendered accent and type usage, chosen logo, social accounts, and industry so the report can disclose where two separate captures agree or conflict.