Skip to content
Supporting proofs / Agent observability

AgentRadar

When several AI agents share a task, a lead agent can report progress while one of its subagents waits for approval. AgentRadar brings their reports into one view so you can follow the work across projects, machines and runtimes.

Follow a task through the team

Find the lead agent

Group the board by project and search for a task. Switch the grouping to machine or runtime to see where the same agents are working.

Inspect its subagents

On a lead agent’s card, open the subagent list. Compare individual tasks and statuses with the counts on the parent card.

Trace a collaboration

Look for a pair of lead agents in the collaboration panel. Use the recent events list to see what each agent reported and when.

What the recording shows

The browser replays a session recorded in AgentRadar v0.4, but all agents, projects and machines in the demo are fictional. The recording plays in a loop. Both languages show the same events. Each status reflects an agent’s own report.

Explore the other supporting proofs →