Installation

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "kansei-link": {
      "command": "npx",
      "args": ["-y", "@kansei-link/mcp-server"]
    }
  }
}

Cursor (.cursor/mcp.json)

{
  "mcpServers": {
    "kansei-link": {
      "command": "npx",
      "args": ["-y", "@kansei-link/mcp-server"]
    }
  }
}

Claude Code (.mcp.json)

{
  "mcpServers": {
    "kansei-link": {
      "command": "npx",
      "args": ["-y", "@kansei-link/mcp-server"]
    }
  }
}

Your First Call

Step 1: Search for a service

search_services({
  "intent": "send invoice to clients",
  "compact": true,
  "limit": 5
})

Returns ranked services matching your intent, with trust scores and agent-readiness status.

Step 2: Get tips before integrating

lookup({
  "service_id": "freee"
})

Returns auth setup, common pitfalls, workarounds, and recent reliability data — everything you need before writing integration code.

Step 3: Report your experience

report({
  "service_id": "freee",
  "success": true,
  "latency_ms": 450,
  "task_type": "create_invoice"
})

Takes 5 seconds. Helps every future agent that uses this service.

That's the entire flow. Three tools, three steps.

Going Deeper

Need more than tips? The lookup tool has 8 modes:

// Full connection guide (auth, endpoints, rate limits)
lookup({ "service_id": "freee", "detail": true })

// Usage analytics (success rates, error patterns)
lookup({ "service_id": "freee", "insights": true })

// Multi-service workflow recipe
lookup({ "goal": "onboard new employee in Japan" })

// Reliability history over time
lookup({ "service_id": "freee", "period": "30d" })

All through a single tool. Mode is auto-detected from your parameters.

Standard Flow Guide — Deep dive into the search-lookup-report workflow