One key. Three ways to use it.

Paste this into ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or the equivalent on Windows. Restart Claude Desktop.

{
  "mcpServers": {
    "renavon": {
      "command": "npx",
      "args": ["-y", "@renavon/mcp-server"],
      "env": {
        "RENAVON_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Once connected, ask Claude things like "What were last weekend's Sha Tin winners?" — it'll write the SQL itself.

Cursor reads MCP servers from .cursor/mcp.json in your project. Add this and reload the window.

{
  "mcpServers": {
    "renavon": {
      "command": "npx",
      "args": ["-y", "@renavon/mcp-server"],
      "env": { "RENAVON_API_KEY": "YOUR_API_KEY" }
    }
  }
}

Cursor's chat then has direct access to the same datasets. Works in agent mode too.

ChatGPT and other autonomous agents (Manus, Devin) can query Renavon over plain HTTP. Give the agent these four facts:

• EndpointPOST https://renavon.com/api/v1/query
• HeaderX-API-KEY: YOUR_API_KEY
• Body{"query": "SELECT ..."}
• Databaserenavon_exports (e.g. hkjc_races, crhk_companies)

Try prompts like "Using my Renavon data, show the top 10 jockeys by win rate this season" or "Chart the odds movement for the favourite in the most recent race".

import requests

API_KEY = "YOUR_API_KEY"
API_BASE = "https://renavon.com"

response = requests.post(
    f"{API_BASE}/api/v1/query",
    headers={"X-API-KEY": API_KEY, "Content-Type": "application/json"},
    json={"query": "SELECT * FROM hkjc_races LIMIT 10"}
)
data = response.json()
# data["columns"], data["data"], data["row_count"], data["truncated"]
for row in data["data"]:
    print(row)

Use pandas.DataFrame(data["data"], columns=[c["name"] for c in data["columns"]]) for a DataFrame.

curl -X POST https://renavon.com/api/v1/query \
  -H "X-API-KEY: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "SELECT * FROM hkjc_races LIMIT 10"}'

Works from any HTTP client — Postman, Insomnia, Bun, fetch in Node 18+, etc.