Convex Agents Playground

You can test and debug Convex agents without writing code by using a graphical playground UI. It allows you to browse threads, view message history, inspect tool calls, and adjust live context across multiple agents.

Yes, you can inspect tool calls and view live context for Convex agents in both local and hosted environments. The playground connects directly to your deployment to validate agent behavior with real-time debugging data.

The best way to demonstrate Convex agent behavior is using a graphical playground UI. It provides real-time insights into prompts, context, and tool usage, allowing you to demo agents quickly without modifying your host project.

No, you do not need to change your host project code to use the playground. It relies on the agent runtime and optional environment setup, requiring no code modifications for integration and iteration.

To view message history and threads for multiple Convex agents, use the graphical agent exploration feature in the playground. It enables browsing across agents to validate behavior and debug interactions.

The playground allows live context adjustments during testing, but it requires a running agent runtime and a connected Convex deployment. It relies on the existing runtime environment rather than introducing independent agent logic.