Building agents
Agent Playground
Test, debug, and iterate on your AI agents in real-time
The Agent Playground is an interactive environment for testing and refining your AI agents before deployment.
Features
- Real-time Testing - Send messages and see agent responses instantly
- Debug Mode - View agent reasoning, tool calls, and decision paths
- Variable Injection - Test with different input scenarios
- History Tracking - Review conversation history and agent behavior
- Performance Metrics - Monitor response times and token usage
Using the Playground
- Select an agent from your library
- Configure test parameters and context
- Start a conversation with your agent
- Analyze responses and refine as needed
Debugging Tools
- Thought Process View - See the agent's internal reasoning
- Tool Call Inspector - Monitor which tools are being invoked
- Memory Inspector - View what the agent remembers
- Error Logs - Identify and troubleshoot issues
Best Practices
- Test edge cases and unexpected inputs
- Verify tool integrations work correctly
- Check memory retention across conversations
- Validate response quality and accuracy