Skip to main content

Connecting Vectorize MCP Server to OpenAI Playground

This guide shows you how to connect your Vectorize MCP agent to OpenAI Playground, enabling you to test and interact with your agent's capabilities directly in OpenAI's development environment.

Prerequisites​

  • An OpenAI account with API access
  • A Vectorize MCP agent already created with at least one tool configured
  • An API key for your Vectorize agent

Setup Instructions​

Step 1: Get Your Agent Credentials​

From your Vectorize agent:

  1. Navigate to your agent's Agent API Keys section
  2. If you don't have an API key yet:
    • Click Create New Key
    • Give it a descriptive name (e.g., "OpenAI Playground Key")
    • Click Create Key
  3. Note down:
    • Your Agent ID from the agent overview page
    • Your API Key - save this securely as you won't be able to see it again!

Step 2: Access OpenAI Playground​

  1. Go to OpenAI Playground
  2. Select your preferred model
  3. Navigate to the Tools section in the playground interface

Step 3: Add Your Vectorize Agent​

OpenAI Playground Tools Section

In the OpenAI Playground actions configuration:

  1. Add New Action
  2. Configure the Action Details:
    • URL: https://agents.vectorize.io/api/agents/YOUR_AGENT_ID
    • Label: Descriptive label
    • Description: Add a description of what your agent does

Replace YOUR_AGENT_ID with your actual agent ID from Step 1.

  1. In the API Key / Auth token section:
    • Enter your Vectorize API key
    • The system will automatically format it as a Bearer token

Step 4: Configure Approval Settings​

OpenAI Playground Approval Settings

Configure Approval Settings:

  • You can choose "Always require approval for all tool calls" for testing
  • Or allow automatic execution for faster interactions

Step 5: Test Your Integration​

  1. Start a conversation in the playground
  2. Ask questions related to your agent's capabilities
  3. The playground will automatically suggest using your tools when relevant
  4. Depending on your approval settings, you may need to approve tool calls

OpenAI Playground Tool Usage

Next Steps​

Once your integration is working in the playground:

  • Test all your agent's tools thoroughly
  • Document any issues or limitations you discover
  • Consider setting up monitoring for production use
  • Explore advanced MCP features and capabilities
  • Plan your production deployment strategy

Your Vectorize MCP agent is now integrated with OpenAI Playground and ready for testing and development!

Was this page helpful?