MCP Integration

Connect Verificate with your development environment using the Model Context Protocol

Cursor IDE Integration

1. Configure MCP Settings

Add Verificate MCP server to your Cursor configuration using your own provider API key (OpenAI, Anthropic/Claude, or Gemini). Only one key is required.

~/.cursor/mcp.json

{
  "mcpServers": {
    "verificate": {
      "command": "node",
      "args": ["/path/to/verificate-mcp-client.js"],
      "env": {
        "MCP_URL": "https://<your-worker>.workers.dev",
        "OPENAI_API_KEY": "sk-... (or ANTHROPIC_API_KEY / GEMINI_API_KEY)"
      }
    }
  }
}

2. Install MCP Server

npm install -g @verificate/mcp-server
# or
npx @verificate/mcp-server install

3. Verify Connection

Restart Cursor and check the MCP connection status in the bottom bar.

Claude Code Integration

1. MCP Server Setup

Configure Claude to use the Verificate MCP Worker with your own provider API key. Only one key is required.

Claude Desktop Configuration

{
  "mcp": {
    "servers": {
      "verificate": {
        "command": "node",
        "args": ["/path/to/verificate-mcp-client.js"],
        "env": {
          "MCP_URL": "https://<your-worker>.workers.dev",
          "OPENAI_API_KEY": "sk-... (or ANTHROPIC_API_KEY / GEMINI_API_KEY)"
        }
      }
    }
  }
}

2. Provider API Key (BYO)

# Choose ONE provider key (do not set all)
export OPENAI_API_KEY="sk-..."      # or
export ANTHROPIC_API_KEY="sk-ant-..." # or
export GEMINI_API_KEY="ai-..."

# Claude Desktop JSON can reference these via the env block in your MCP config.

Cloudflare Light Connection

1. Worker Deployment

Deploy the Verificate MCP bridge as a Cloudflare Worker. No server-side secrets are required; users provide their own API keys per request.

# Install Wrangler CLI
npm install -g wrangler

# Deploy MCP bridge
wrangler deploy --name verificate-mcp-bridge

2. Connection URL

Use this endpoint in your MCP client:

https://verificate-mcp-bridge.your-subdomain.workers.dev

3. Configuration

{
  "mcpServers": {
    "verificate-cloud": {
      "command": "node",
      "args": ["/path/to/verificate-mcp-client.js"],
      "env": {
        "MCP_URL": "https://verificate-mcp-bridge.your-subdomain.workers.dev",
        "OPENAI_API_KEY": "sk-... (or ANTHROPIC_API_KEY / GEMINI_API_KEY)"
      }
    }
  }
}

Direct MCP Connection

Local Development

Run Verificate MCP server locally:

# Clone the MCP server
git clone https://github.com/verificate-ai/mcp-server.git
cd mcp-server

# Install dependencies
npm install

# Start the server
npm start -- --port 8004

# Server runs on http://localhost:8004

Connection Test

# Test MCP connection
curl -X POST http://localhost:8004/validate-ai-output \
  -H "Content-Type: application/json" \
  -d '{"ai_output": "test validation"}'

Need Help with Integration?

Our team is here to help you get connected

Schedule Demo

Connect Your IDE via Serverless MCP

Use the gateway below to verify the Cloudflare Worker is online, then point Cursor, Claude Code, or Windsurf AI to the Worker URL.

Configuration Snippets

Add your API key directly in your IDE's JSON config. You only need one provider (OpenAI, Anthropic/Claude, or Google Gemini) — include just the key you plan to use.

Cursor IDE (~/.cursor/mcp.json)

{
  "mcpServers": {
    "verificate": {
      "command": "node",
      "args": ["/path/to/verificate-mcp-client.js"],
      "env": {
        "MCP_URL": "https://<your-worker>.workers.dev",
        "OPENAI_API_KEY": "sk-... (optional)",
        "ANTHROPIC_API_KEY": "sk-ant-... (optional)",
        "GEMINI_API_KEY": "ai-... (optional)"
      }
    }
  }
}

Tip: Only one of the keys above is required. Use the provider you prefer.

Claude Code (Desktop config)

{
  "mcp": {
    "servers": {
      "verificate": {
        "command": "node",
        "args": ["/path/to/verificate-mcp-client.js"],
        "env": {
          "MCP_URL": "https://<your-worker>.workers.dev",
          "ANTHROPIC_API_KEY": "sk-ant-... (or OPENAI_API_KEY / GEMINI_API_KEY)"
        }
      }
    }
  }
}

Note: Provide exactly one key matching your chosen provider.

Windsurf AI

# Environment variables (recommended)
MCP_URL=https://<your-worker>.workers.dev
OPENAI_API_KEY=sk-...   # or
ANTHROPIC_API_KEY=sk-ant-...   # or
GEMINI_API_KEY=ai-...

# If your Windsurf MCP plugin supports JSON config with env, set the same keys there.

Use any single provider key. Keys are never stored on our servers.