Datasette API

{
  "service_type": "platform",
  "base_url": "https://datasette.local",
  "auth_method": "none",
  "auth_config": {},
  "endpoints": [
    {
      "path": "/.json",
      "method": "GET",
      "description": "Returns metadata about the Datasette instance and available databases in JSON format"
    },
    {
      "path": "/{database}.json",
      "method": "GET",
      "description": "Returns metadata about a specific database"
    },
    {
      "path": "/{database}/{table}.json",
      "method": "GET",
      "description": "Returns rows from a table as JSON, supports pagination via _size and _next parameters"
    },
    {
      "path": "/{database}.json",
      "method": "GET",
      "description": "Execute arbitrary SQL queries via the ?sql= parameter (read-only queries)"
    },
    {
      "path": "/{database}/{table}/{pk}.json",
      "method": "GET",
      "description": "Returns a single row by primary key as JSON"
    },
    {
      "path": "/-/metadata.json",
      "method": "GET",
      "description": "Returns instance metadata including license, source, and description"
    },
    {
      "path": "/-/versions.json",
      "method": "GET",
      "description": "Returns versions of Datasette, SQLite, and Python"
    },
    {
      "path": "/-/plugins.json",
      "method": "GET",
      "description": "Lists installed Datasette plugins"
    },
    {
      "path": "/-/settings.json",
      "method": "GET",
      "description": "Returns Datasette settings"
    },
    {
      "path": "/-/threads.json",
      "method": "GET",
      "description": "Returns information about active threads"
    }
  ],
  "pricing_model": {
    "type": "free",
    "details": {
      "note": "Datasette is open-source software (Apache 2.0). Hosting is self-managed; costs depend on infrastructure. Datasette Cloud offers a paid hosted service with a freemium tier."
    }
  },
  "rate_limits": {},
  "capabilities": [
    "Execute read-only SQL queries against SQLite databases",
    "Return query results as JSON",
    "Explore database schema and metadata",
    "Paginate and filter table data",
    "Plugin ecosystem for custom functionality",
    "Publish and share data as a web interface",
    "Support for full-text search",
    "Faceted browsing and filtering",
    "CSV and other export formats",
    "Authentication and permissions (via plugins like datasette-auth-tokens)"
  ],
  "raw_analysis": "Datasette is an open-source tool for exploring and publishing data, created by Simon Willison. It provides an instant JSON API for any SQLite database, making it ideal for data journalists, researchers, and developers who need to query and share datasets. The platform is mature, with active development since 2017 and a rich plugin ecosystem. For a local instance at https://datasette.local, the API is typically unauthenticated unless configured otherwise. Key capabilities include executing arbitrary read-only SQL queries via the ?sql= parameter, retrieving table rows with pagination, and accessing metadata endpoints. Datasette also supports full-text search, faceted filtering, and various output formats. While the core is free and open-source, Datasette Cloud provides a hosted, paid alternative. Integration options include plugins for authentication, custom visualizations, and data import/export. The API is RESTful and returns JSON by default when .json is appended to URLs or when using the ?_format=json parameter."
}

0/3 tests passed

{
  "overall": 52,
  "dimensions": {
    "token_efficiency": 7.5,
    "first_try_success": 5.5,
    "response_parseability": 8.5,
    "error_clarity": 7.0,
    "doc_quality": 7.5,
    "auth_simplicity": 6.0,
    "latency": 1.0,
    "consistency": 4.0
  },
  "pricing_normalized": {
    "model": "open_source_free",
    "self_hosted": "free (Apache 2.0) + infrastructure costs",
    "managed": "Datasette Cloud (freemium + paid tiers)",
    "agent_friendliness": "self-hosted with token auth is agent-friendly; Datasette Cloud requires account setup"
  },
  "issues": [
    "All three live checks failed with DNS resolution errors (Name or service not known) — the primary domain could not be reached from the test environment, making uptime/latency/robots.txt/llms.txt unverifiable.",
    "No hosted endpoint verified — agents must either self-host or onboard to Datasette Cloud, adding setup friction before first query.",
    "No llms.txt detected, so agent-oriented discovery of capabilities is unavailable from the primary domain.",
    "Auth is plugin-dependent (e.g. datasette-auth-tokens); no built-in magic-link or verified-identity flow, so agents can't self-onboard out of the box.",
    "Latency scored low (1.0) because live response time is unmeasurable due to DNS failure; the measured ~75-95ms are on failed lookups, not real requests."
  ],
  "recommendations": [
    "Verify the canonical deployment URL / ensure DNS is stable so uptime, latency, and status pages can be measured; this is the biggest blocker to scoring higher.",
    "Publish an llms.txt or agent-discovery doc at the root describing available endpoints, SQL surface, and auth tokens.",
    "Document a clear MCP or HTTP API path with example queries so agents achieve first-try success without reading source.",
    "Provide a magic-link or token-issuance flow for Datasette Cloud so agents can onboard without a human handshake.",
    "Surface a public status page / SLOs to back up consistency scoring.",
    "Ship an OpenAPI/JSON-schema spec for the query and metadata endpoints to boost parseability fidelity."
  ]
}