One instrument. One useful task.
See who does what
Step 1: Lift the illusion
You set the scope, agents propose changes, and you inspect the result before deciding what ships.
Still frames · every step is available
“See who does what.”
You set the scope, agents propose changes, and you inspect the result before deciding what ships.
See who does what
Choose the task, workspace and permitted scope.
You set the scope, agents propose changes, and you inspect the result before deciding what ships. Start here: Demonstration → inspect the mechanism.
Identify the task scope and the decision you still need to make.
This is an explanatory diagram, not live app telemetry.
Before you start
Launch a supported agent inside Code Overlord so it receives the app's local MCP tools. The tools let it discover available agents, create sessions, send input and read results. A separately launched external chat does not automatically have the same tool access.
Illustrated explanation · The scene explains the workflow. Follow the guide to perform it in the app.
About this demonstration
Instructions reviewed against product 1.5.119. Authored illustration; no application execution is claimed.
