Reports
The Reports tab gives a per-agent view of throughput, delegation patterns, execution time, and skill activity over time.
What Reports shows
| Metric | What it measures |
|---|---|
| Tasks completed | Total tasks each agent finished within the selected period, grouped by status. |
| Delegation count | How many times each agent was delegated work via agent-send, broken down by source agent. |
| Average execution time | Mean time from task creation to completion per agent, useful for spotting bottlenecks. |
| Skill usage | Which skills each agent invoked most: browse, memory, github, cron, home-assistant, and so on. |
How to read the data
| Signal | What it may mean |
|---|---|
| High delegation count, low own-completions | The agent is acting as a coordinator. Normal for main; worth reviewing if a specialist is over-delegating. |
| Long average execution time on one agent | The agent may be waiting on external tools, blocked tasks, or receiving under-scoped delegation briefs. |
| Skill usage dominated by browse | The agent is research-heavy. Consider whether memory could cache repeated lookups. |
| Low throughput across all agents | Might reflect a slow period, or missions with many blocked tasks awaiting user input. |
Scope and period
Reports aggregate data from the team activity log written during normal agent turns. The dashboard lets you filter by agent and time window. Historical data is retained as long as the activity log exists in your state directory.
Reports and missions
Task completions in Reports correspond to Tasks inside Missions. If you delete a mission, its associated activity entries are pruned from the log. Reports for that mission's work will no longer appear in future queries. Active and paused missions continue to accumulate entries as agents work on them.
Related docs
See Missions for how tasks are created and tracked, and Skills for the full list of skill names that appear in usage data.