Oraclaw

No local installation is required. Add the MCP server to your Claude Desktop config at ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%/Claude/claude_desktop_config.json (Windows). The npx command pulls the latest package automatically.

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

To unlock the six premium tools (CMA-ES, HiGHS solver, graph analytics, VaR/CVaR risk, ARIMA forecasting, anomaly detection), add your ORACLAW_API_KEY to the environment block. Without a key, the 11 free-tier tools work with 25 calls/day.

{
  "mcpServers": {
    "oraclaw": {
      "command": "npx",
      "args": ["-y", "@oraclaw/mcp-server"],
      "env": {
        "ORACLAW_API_KEY": "your-oraclaw-api-key-here"
      }
    }
  }
}

Quit and relaunch Claude Desktop. The Oraclaw tools will appear in Claude's tool list. Ask Claude to describe the available decision intelligence tools to verify the connection.

Ask Claude to run a multi-armed bandit using UCB1 to choose between options. Claude will call the optimize_bandit tool with your arm data and return the recommended next action with confidence scores.

# Example REST API call (for testing without Claude Desktop)
curl -X POST https://oraclaw-api.onrender.com/api/v1/optimize/bandit \
  -H 'Content-Type: application/json' \
  -d '{
    "arms": [
      {"id": "A", "name": "Option A", "pulls": 10, "totalReward": 7},
      {"id": "B", "name": "Option B", "pulls": 8, "totalReward": 4},
      {"id": "C", "name": "Option C", "pulls": 5, "totalReward": 3}
    ],
    "algorithm": "ucb1"
  }'

Ask Claude to run a Monte Carlo simulation to model uncertainty in a decision or forecast. The simulate_montecarlo tool supports six distributions: normal, uniform, lognormal, exponential, triangular, and beta.