Activity

What it is

The Activity page is the execution hub for a program. It provides a dashboard view of all AI agent executions, an execution trace viewer for debugging, coverage detail showing which requirements have been touched by agents, and a manual skill execution modal.

Why it matters

Visibility into AI agent activity is critical for trust. Activity lets you see what agents did, how long they took, what they produced, and which parts of the program have (or have not) received AI attention.

Key concepts

View states: dashboard, trace, coverage — three views accessible from the activity page
Activity dashboard: ActivityDashboard shows a summary of recent executions with status indicators and search
Activity trace: ActivityTrace renders a detailed trace of a single execution with input/output
Trace detail sections: TraceDetailSections breaks down execution steps within a trace
Trace group: Groups related trace entries for hierarchical display
Coverage detail: CoverageDetail shows which requirements have been covered by agent executions, grouped by workstream
Execute skill modal: ExecuteSkillModal lets you manually trigger a skill execution from the activity page
Metric card: MetricCard displays execution-level metrics (duration, tokens, cost)
Execution output: ExecutionOutput renders the raw or formatted output from an agent execution
Search: Debounced search across execution summaries

How to use it

Navigate to [Program] > Activity.
View the Dashboard for a summary of recent executions.
Search for specific executions using the search bar.
Click an execution to open the Trace view with full input/output detail.
Switch to the Coverage view to see which requirements have been touched by agents.
Click Execute Skill to manually run a skill against selected requirements.