Skip to content

Getting Started

Get up and running with Prediction Markets MCP in under 5 minutes.

Prerequisites

  • Node.js v18+ (for npx) or Bun v1.0+ (for bunx)
  • An MCP-compatible client (see supported clients below)

Installation

This server is published on npm. No cloning or building required—just configure your MCP client to run it via npx.

Register with Your MCP Client

Choose your client and follow the setup instructions.

Configuration may vary

MCP client configuration varies by tool—some support project-level, user-level, or both. The examples below show typical setups. Check your client's official documentation for the latest details.

Restart often required

Most MCP clients load servers at startup. After adding or updating configuration, you may need to restart your client (or use a refresh button, depending on the client) to load changes.

Add to .mcp.json in your project or ~/.claude.json for global access:

{
  "mcpServers": {
    "prediction-markets": {
      "command": "npx",
      "args": ["-y", "prediction-mcp"],
      "env": {
        "KALSHI_API_KEY": "your-api-key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

Using Bun? Replace "command": "npx", "args": ["-y", "prediction-mcp"] with "command": "bunx", "args": ["prediction-mcp"].

Edit claude_desktop_config.json at the path for your OS:

OS Path
macOS ~/Library/Application Support/Claude/claude_desktop_config.json
Windows %APPDATA%\Claude\claude_desktop_config.json

Quick access

In Claude Desktop: Settings → Developer → Edit Config

Add the following configuration:

{
  "mcpServers": {
    "prediction-markets": {
      "command": "npx",
      "args": ["-y", "prediction-mcp"],
      "env": {
        "KALSHI_API_KEY": "your-api-key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

VS Code has native MCP support.

Option 1: Workspace configuration

Create .vscode/mcp.json in your project:

{
  "servers": {
    "prediction-markets": {
      "command": "npx",
      "args": ["-y", "prediction-mcp"],
      "env": {
        "KALSHI_API_KEY": "your-api-key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

Option 2: Command Palette

  1. Open Command Palette (Ctrl/Cmd + Shift + P)
  2. Run MCP: Add Server
  3. Enter the server configuration

More info

See VS Code MCP documentation for advanced features like input prompts and HTTP transport.

Create .cursor/mcp.json in your project root, or ~/.cursor/mcp.json for global access:

{
  "mcpServers": {
    "prediction-markets": {
      "command": "npx",
      "args": ["-y", "prediction-mcp"],
      "env": {
        "KALSHI_API_KEY": "your-api-key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

Alternative: Settings UI

  1. Go to File → Preferences → Cursor Settings
  2. Select MCP
  3. Click Add Custom MCP

More info

See Cursor MCP documentation for environment variable interpolation and advanced configuration.

Edit the MCP config file at ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "prediction-markets": {
      "command": "npx",
      "args": ["-y", "prediction-mcp"],
      "env": {
        "KALSHI_API_KEY": "your-api-key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

Alternative: Plugin Store

  1. Click the Plugins icon in the Cascade panel
  2. Or go to Windsurf Settings → Cascade → Plugins

More info

See Windsurf MCP documentation for team administration and whitelisting.

Add to your Zed settings.json (Preferences → Settings):

{
  "context_servers": {
    "prediction-markets": {
      "command": "npx",
      "args": ["-y", "prediction-mcp"],
      "env": {
        "KALSHI_API_KEY": "your-api-key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

Check the server status in the Agent Panel settings—a green indicator means it's active.

More info

See Zed MCP documentation for extension-based MCP servers.

Create opencode.jsonc in your project root:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "prediction-markets": {
      "type": "local",
      "command": ["npx", "-y", "prediction-mcp"],
      "enabled": true,
      "environment": {
        "KALSHI_API_KEY": "your-api-key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

More info

See OpenCode MCP documentation for remote servers and OAuth configuration.

Verify It Works

Ask your AI assistant:

"List open markets on Kalshi about the Federal Reserve"

Or for Polymarket (no credentials needed):

"What are the top markets on Polymarket right now?"

If the server is working, you'll see market data in the response. If not, see Troubleshooting.

Credentials

Polymarket

No credentials required. All Polymarket tools work out of the box—read operations are public.

Kalshi

Kalshi requires API credentials:

  1. Create an account at kalshi.com
  2. Go to kalshi.com/account/profile
  3. Generate an API key and download your private key

Demo environment

Kalshi offers a demo environment with mock funds for testing. Demo credentials are separate from production—create a demo account at demo.kalshi.co first, then set KALSHI_USE_DEMO=true in your configuration.

See Configuration for all environment variables.

Next Steps