>_ The Manifest

The agent worked in your demo. Then a real user phrased the same request differently and it fell over. That is not a disaster, it is a debugging session. You just need the right tools open. Watch the walkthrough, then skim the recap below.

▶ Testing and debugging your Copilot Chat declarative agent on the Microsoft 365 & Power Platform Community channel. Open on YouTube.

What you’ll see

Testing a declarative agent means watching what it actually does, not just what it says. This walkthrough opens the developer tooling that shows the agent’s reasoning and where it goes wrong.

  • Developer mode and debug cards to see the agent’s decisions in real time
  • Capability tracing to confirm which knowledge, plugin, or action fired
  • App versioning and environment files to keep test and production separate
  • Practical troubleshooting for APIs, MCP servers, grounding, and instructions

The takeaway: trust comes from evidence. Before you claim an agent got better, prove it. This is where evals and disciplined testing start.

Resources

Have questions or want to share what you're building? Connect with me on LinkedIn or check out more on The Manifest.