Firecrawl

{
  "score": 98,
  "tier": "Agent Native",
  "history": [
    {
      "date": "2026-06-07",
      "score": 98,
      "tier": "Agent Native",
      "summary": "Initial evaluation. Firecrawl is highly agent-native: bearer-token API auth, AI onboarding docs, llms.txt and llms-full.txt, hosted and local MCP servers, CLI and agent skills, strong SDK coverage, webhooks with retries, monitor events, rate-limit documentation, public status, and a structured error catalog. The score is held below a perfect mark because default account/key creation still appears dashboard-oriented unless a platform can use the WorkOS ID-JAG path, and public API-key scoping/audit-log details were not found."
    }
  ],
  "breakdown": [
    {
      "index": "01",
      "criterion": "Machine Authentication",
      "score": 20,
      "max": 20,
      "note": "The v2 API uses bearer-token authentication via the Authorization header and documented FIRECRAWL_API_KEY environment variable flows for SDKs, CLI, and MCP. No browser redirect or OAuth loop is required once an API key exists."
    },
    {
      "index": "02",
      "criterion": "Autonomous Onboarding",
      "score": 12,
      "max": 20,
      "note": "Firecrawl publishes AI-agent onboarding links, a free monthly credit allowance, CLI/browser auth setup, and an auth.md path for platforms that can mint a WorkOS ID-JAG. The ordinary API-key path still points agents to the dashboard, so fully autonomous signup/key issuance is not universally documented."
    },
    {
      "index": "03",
      "criterion": "MCP Support",
      "score": 20,
      "max": 20,
      "note": "Official MCP support is first-class via a hosted streamable HTTP endpoint and the firecrawl-mcp npm package. The MCP server is open source and covers search, scrape, interact, crawl, map, extract, agent, browser sessions, cloud, and self-hosted use."
    },
    {
      "index": "04",
      "criterion": "API Documentation Quality",
      "score": 10,
      "max": 10,
      "note": "Documentation includes llms.txt and llms-full.txt, a v2 API reference, endpoint pages for search/scrape/parse/map/crawl/monitor/account usage, SDK guides for Node.js, Python, Go, Rust, Elixir, Java, .NET, PHP, and Ruby, cookbook examples, integration guides, and public GitHub docs sources."
    },
    {
      "index": "05",
      "criterion": "Real-Time Event Delivery",
      "score": 15,
      "max": 15,
      "note": "Webhooks support real-time started/page/completed/failed-style events for crawl, batch scrape, extract, agent jobs, and monitor checks. Delivery requires HTTPS, includes configurable metadata/events, and retries failed deliveries after 1, 5, and 15 minutes."
    },
    {
      "index": "06",
      "criterion": "Structured Error Responses",
      "score": 9,
      "max": 10,
      "note": "Firecrawl documents a consistent JSON error shape with success=false, error, optional details/code, an error catalog covering common HTTP statuses, retryability, remediation, Retry-After handling for 429s, and copy-paste backoff snippets. The catalog is explicitly non-exhaustive, so it misses full marks."
    },
    {
      "index": "07",
      "criterion": "Security Posture",
      "score": 8,
      "max": 10,
      "note": "Security posture is strong: bearer API keys, HTTPS webhook requirements, HMAC-SHA256 webhook signatures, timing-safe verification guidance, documented rate/concurrency limits, public status monitoring, robots.txt respect, and public SOC 2 Type II trust signals. Public audit logs and granular API-key scopes were not found."
    },
    {
      "index": "08",
      "criterion": "Response Consistency",
      "score": 4,
      "max": 5,
      "note": "The API is documented under /v2, GitHub shows active versioned releases, and Firecrawl publishes a high-cadence changelog/blog. A formal deprecation or backward-compatibility policy was not found in reviewed public materials."
    }
  ]
}

{
  "score": 93,
  "summary": "Very strong market confidence from broad developer adoption, named enterprise users, Series A backing, a large open-source community, mature documentation, status visibility, and public SOC 2 Type II trust signals.",
  "signals": [
    {
      "criterion": "Adoption",
      "score": 20,
      "max": 20,
      "note": "Firecrawl publicly claims 1.25M+ developers, 150,000+ companies, 8B+ pages fetched, and named users including Apple, Canva, Shopify, Zapier, Lovable, Replit, Sierra, and Gamma."
    },
    {
      "criterion": "Funding & Backing",
      "score": 17,
      "max": 20,
      "note": "Firecrawl announced a $14.5M Series A led by Nexus Venture Partners in August 2025, bringing total funding to $16.2M with participation from Y Combinator, Zapier, Shopify CEO Tobias Lütke, and other angels."
    },
    {
      "criterion": "Community & Ecosystem",
      "score": 20,
      "max": 20,
      "note": "The open-source repository shows about 130k GitHub stars and 7.7k forks, AGPL-3.0 core licensing, 34 releases, multi-language SDKs, official MCP, CLI, agent skills, workflows, Vercel AI SDK tools, and broad framework integrations."
    },
    {
      "criterion": "Operational Maturity",
      "score": 18,
      "max": 20,
      "note": "Firecrawl publishes detailed docs, pricing, rate limits, a public status page, changelog/blog release cadence, hosted API, self-hosting docs, webhooks, queue status, and enterprise paths. The main maturity caveat is that rapid product expansion creates more surface area to keep stable."
    },
    {
      "criterion": "Security & Compliance",
      "score": 18,
      "max": 20,
      "note": "Public materials reference SOC 2 Type II, API key authentication, HTTPS webhook endpoints, HMAC-SHA256 webhook signatures, rate and concurrency limits, Retry-After handling, status monitoring, robots.txt respect, and zero-data-retention positioning. Public audit-log and API-key scoping details were not found."
    }
  ]
}