Agent Standup Beta

The same daily digest yeaboi produces for your team, for the agents working beside it: which projects they worked, what landed, and what is still open.

Beta. The Agents family is new, and every number it shows is computed from telemetry your agents happen to leave behind. Treat it as an indicator, not an audit.

The window

By default the digest reaches back to the start of the previous working day, which is the same rule the human standup uses: a Monday run covers Friday, so a weekend never hides a Saturday night's agent work. Pass --days N to look back a fixed number of days instead.

Sources

Local session rollups are included by default. On top of those, yeaboi can scan your trackers for agent-authored commits and pull requests — GitHub and Azure DevOps — so the digest covers work that left the machine as well as work that happened on it. Tracker scanning is best-effort: it is skipped with a warning rather than failing the run.

yeaboi agents standup                          # local + both trackers
yeaboi agents standup --tracker-sources        # local only
yeaboi agents standup --tracker-sources github
yeaboi agents standup --github-owners my-org
yeaboi agents standup --no-local-sessions      # trackers only

--no-local-sessions skips the local scan entirely rather than running it and discarding the result, and the digest says so in its coverage notes. Use it wherever ~/.claude is not the machine whose work you are reporting on — a CI job or a scheduled cloud run, where the only sessions on disk are that job's own. Without it such a run reports itself as the team's agent activity.

Delivery

--deliver posts the digest to your configured Slack webhook and raises a desktop notification. Delivery never raises — a webhook that is down becomes a warning on the report, not a failed run, because the digest itself is still worth having.

What the digest answers

Highlights and points needing attention are drawn from the computed rows — sessions worked, projects touched, commits and PRs authored, and threads left open — and the narrative is the single LLM call over them. As with the other Agents modes, the model writes prose and never numbers.

From MCP

agents_standup_run builds a digest; agents_standup_history lists stored ones.