Perfloop API

{
  "service_type": "platform",
  "base_url": "https://app.perfloop.ai",
  "auth_method": "unknown",
  "auth_config": {},
  "endpoints": [
    {
      "path": "/mcp",
      "method": "unknown",
      "description": "Model Context Protocol endpoint referenced at app.perfloop.ai/mcp; likely used to expose Perfloop tooling/capabilities to LLM-based clients."
    }
  ],
  "pricing_model": {
    "type": "unknown",
    "details": {
      "note": "No public pricing information is available for app.perfloop.ai or Perfloop API."
    }
  },
  "rate_limits": {
    "documented": false,
    "note": "No public rate limit documentation found."
  },
  "capabilities": [
    "MCP (Model Context Protocol) integration endpoint",
    "AI-related tooling (inferred from category 'ai')",
    "Unknown core product functionality — public documentation not found",
    "Likely intended for integration with LLM/AI client environments via /mcp"
  ],
  "raw_analysis": "Perfloop API was auto-discovered via a benchmark URL submission with the single URL https://app.perfloop.ai/mcp and categorized under 'ai'. The '/mcp' path strongly suggests a Model Context Protocol endpoint rather than a conventional public REST API; MCP servers typically expose tool/function capabilities to LLM clients (e.g., Claude Desktop, IDE agents) rather than standard CRUD REST resources. The base application appears to live at app.perfloop.ai, which is consistent with a hosted SaaS web app. There is insufficient public information to determine what Perfloop does beyond its AI categorization — no visible homepage description, docs, pricing page, or API reference was provided. Who it's for: likely AI developers or teams wanting to connect Perfloop functionality into an agent workflow via MCP. Maturity: undeterminable; the presence of an app subdomain and an MCP endpoint suggests it is at least in an early-to-mid stage product with modern AI integrations, but no evidence of a stable public REST API, versioning, or SLA. Integrations: the primary integration surface appears to be MCP, implying compatibility with MCP-capable AI clients and tool-calling frameworks. Authentication, rate limits, and pricing are all unknown and would need to be confirmed by inspecting the MCP server handshake (e.g., MCP 'initialize' and 'tools/list' responses) or any available documentation at the root domain. Recommendation: treat as a platform with an MCP-first interface; probe /mcp with an MCP client to enumerate available tools and auth requirements before building any integration."
}

1/3 tests passed

{
  "overall": 38,
  "dimensions": {
    "token_efficiency": 3.0,
    "first_try_success": 2.0,
    "response_parseability": 3.0,
    "error_clarity": 2.0,
    "doc_quality": 1.0,
    "auth_simplicity": 3.0,
    "latency": 6.0,
    "consistency": 4.0
  },
  "pricing_normalized": {
    "model": "unknown",
    "note": "No public pricing information found for app.perfloop.ai or Perfloop API."
  },
  "issues": [
    "No public documentation, pricing, or product description — cannot explain value prop to users",
    "No llms.txt or robots.txt (both 404) — poor agent-readability signals",
    "Website is a client-rendered SPA that reveals almost nothing without JS execution; static fetch yields only shell title",
    "Unknown core functionality despite 'AI' category; MCP endpoint present but undocumented",
    "Onboarding, auth, and signup flows are undiscoverable without human browsing"
  ],
  "recommendations": [
    "Publish an llms.txt and robots.txt describing the product, MCP endpoint, and intended use",
    "Add a public docs/landing page with plain-text value prop, pricing, and getting-started steps",
    "Document the MCP endpoint contract (transport, auth, tools exposed) for agent consumption",
    "Provide server-side rendered or static marketing content so agents can parse capability without JS",
    "Expose a status page and auth details to improve reliability perception"
  ]
}