Build with MCP Analytics

// Add to Cursor or Windsurf MCP settings
// Settings > MCP Servers > Add Server

{
  "mcpServers": {
    "mcp-analytics": {
      "command": "npx",
      "args": ["@anthropic/mcp-analytics"],
      "env": {
        "MCP_AUTH_SERVER": "https://api.mcpanalytics.ai"
      }
    }
  }
}

// OAuth2 login opens in browser on first use
// After auth, tools are available immediately

# 1. Upload a dataset
curl -X POST https://api.mcpanalytics.ai/datasets/upload \
  -H "Authorization: Bearer $TOKEN" \
  -F "file=@sales_data.csv"

# Response: { "dataset_id": "ds_abc123...", "columns": [...] }

# 2. Discover compatible analysis tools
curl -X POST https://api.mcpanalytics.ai/discover/tools \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"dataset_id": "ds_abc123", "query": "revenue trends"}'

# 3. Run the analysis
curl -X POST https://api.mcpanalytics.ai/tools/run \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"tool": "time_series_arima", "dataset_id": "ds_abc123",
       "column_mapping": {"date": "order_date", "value": "revenue"}}'

# Python — using requests
import requests

API = "https://api.mcpanalytics.ai"
headers = {"Authorization": f"Bearer {token}"}

# Upload dataset
upload = requests.post(
    f"{API}/datasets/upload",
    headers=headers,
    files={"file": open("sales_data.csv", "rb")}
)
dataset_id = upload.json()["dataset_id"]

# Discover tools for this dataset
tools = requests.post(
    f"{API}/discover/tools",
    headers=headers,
    json={"dataset_id": dataset_id, "query": "revenue forecast"}
).json()

# Run the top-matched analysis
result = requests.post(
    f"{API}/tools/run",
    headers=headers,
    json={
        "tool": tools["matches"][0]["tool_name"],
        "dataset_id": dataset_id,
        "column_mapping": {"date": "order_date", "value": "revenue"}
    }
).json()

# result["report_url"] → interactive HTML report
# result["cards"] → individual chart/table components