I parsed 10 years of Japanese corporate filings

{
  "service_type": "platform",
  "base_url": "https://axiora.dev",
  "auth_method": "api_key",
  "auth_config": {
    "header": "X-API-Key",
    "signup_url": "https://axiora.dev/en/pricing"
  },
  "endpoints": [
    {
      "path": "/api.md",
      "method": "GET",
      "description": "Machine-readable API overview"
    },
    {
      "path": "/en/docs/reference",
      "method": "GET",
      "description": "REST API reference"
    },
    {
      "path": "/en/docs/quickstart",
      "method": "GET",
      "description": "Quickstart guide"
    },
    {
      "path": "/en/docs/mcp",
      "method": "GET",
      "description": "MCP server documentation"
    },
    {
      "path": "/openapi.json",
      "method": "GET",
      "description": "OpenAPI JSON specification"
    },
    {
      "path": "/pricing.md",
      "method": "GET",
      "description": "Pricing for AI buying agents"
    },
    {
      "path": "/en/pricing",
      "method": "GET",
      "description": "Public pricing page"
    }
  ],
  "pricing_model": {
    "type": "subscription",
    "details": {
      "free_tier": true,
      "has_subscription": true,
      "note": "Sign up for a free key; paid plans likely based on usage or tier."
    }
  },
  "rate_limits": {},
  "capabilities": [
    "REST API for Japanese financial data",
    "MCP server for AI agents (41 read-only tools)",
    "Source-linked data with EDINET doc_id provenance",
    "Parsed financial statements (JGAAP, IFRS, US GAAP)",
    "Ownership trajectories and large-shareholding movements",
    "Governance data",
    "English translations of filings",
    "Entity resolution across Japanese scripts",
    "Archival of EDINET filings beyond viewing window",
    "Data from 4,125 listed companies",
    "10 years of historical filings",
    "Python SDK",
    "AI agent integration (Claude, ChatGPT)",
    "Reproducible query results via as_of tokens",
    "Comparison pages vs Bloomberg, FactSet, Toyo Keizai, EDINET, EDINETDB, J-Quants, QUICK"
  ],
  "raw_analysis": "Axiora is a niche, developer-first platform that transforms Japan's official EDINET XBRL filings into structured, source-linked financial data, ownership time-series, and governance information. It serves 4,125 listed companies with 10 years of historical data, covering financial statements (JGAAP, IFRS, US GAAP), large-shareholding movements, and governance details. The platform positions itself as a focused alternative to broad enterprise terminals like Bloomberg and FactSet, emphasizing provenance and AI-agent accessibility. It offers a REST API with an OpenAPI spec, a Python SDK, and an MCP server with 41 read-only tools, enabling AI agents like Claude and ChatGPT to query financials with verifiable source links (doc_id). Key differentiators include source traceability (every row links to the EDINET filing), reproducibility (as_of tokens for bit-identical replay), and comprehensive entity resolution across Japanese scripts. The API appears to be production-ready, with free sign-up for a key and a subscription pricing model. It provides detailed documentation, quickstart guides, and comparison pages. The service is mature enough to have a public OpenAPI JSON, an MCP server, and a strong focus on developer experience and auditability. Integrations extend to AI agent frameworks (LangChain is mentioned in blogs), and it handles the archiving burden of EDINET's viewing window. The target audience includes AI agents, funds, developers, and data product builders requiring auditable Japanese disclosure data. Rate limits are not explicitly documented in the provided text, suggesting they may be defined in the full API docs. Overall, the platform is a well-structured, high-fidelity data infrastructure service specializing in Japanese corporate filings, with a strong emphasis on verifiability and AI-readiness."
}

3/3 tests passed

```json
{
  "overall": 82,
  "dimensions": {
    "token_efficiency": 9.0,
    "first_try_success": 8.0,
    "response_parseability": 9.5,
    "error_clarity": 8.0,
    "doc_quality": 8.5,
    "auth_simplicity": 8.5,
    "latency": 6.0,
    "consistency": 8.0
  },
  "pricing_normalized": {
    "free_tier": true,
    "subscription_model": "usage-based tiers",
    "entry_barrier": "free API key signup"
  },
  "issues": [
    "Root website latency at 1169ms is on the slower side, though within acceptable range",
    "Pricing details are vague — 'paid plans likely based on usage or tier' suggests unclear cost transparency for agents to relay to users",
    "Requires signup for API key, adding one step before first productive call",
    "No explicit error documentation visible in the provided capabilities summary"
  ],
  "recommendations": [
    "Add a clear, structured pricing page with exact per-request costs to facilitate agent-to-user cost estimation",
    "Provide an explicit 'Getting Started in 5 minutes' guide with a single curl command example",
    "Document error codes and rate limits prominently in the MCP tool descriptions",
    "Optimize root page CDN caching to reduce latency below 500ms for faster agent probes"
  ]
}
```