MCP Protocol

Attribution.ai exposes a production MCP JSON-RPC endpoint for connectors and agent runtimes. For end-user setup, start at AI Connectors.

How It Works

1. Your client authenticates with OAuth access token or bearer API key
2. The client negotiates protocol support with initialize
3. Tool discovery and execution happen through JSON-RPC methods
4. Responses return MCP content payloads your agent can parse or summarize

Server Details

Tools (16)

• discover — Data Discovery & Store Context
• explain — Documentation & Methodology
• ask — Ask a Question
• performance — Store & Channel Performance
• campaigns — Campaign & Creative Performance
• trends — Revenue & Order Trends
• customers — Customer Analytics
• surveys — Survey Responses
• products — Products & Coupons
• optimization — Budget Optimization
• experiments — Incrementality Testing
• insights — AI Insights
• manage — Manage Store Actions
• influencers — Influencer Performance
• compare — Period Comparison
• diagnose — Connection & Data Health Check

View the complete tool reference →

Setup by Client

{
  "mcpServers": {
    "attribution-ai": {
      "url": "https://attribution.ai/api/mcp/rpc",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Supported Methods

• initialize
• tools/list and tools/call
• resources/list and resources/read
• prompts/list and prompts/get
• notifications/initialized

Best Practices

• Use specific date ranges and channels to get faster, more relevant results
• Start with default detail_level=summary; request detail_level=analyst only when you need more depth
• Handle 401 and 429 responses explicitly in your client runtime
• Prefer structured parsing over prompt-only interpretation for automation