What you can type at teamtester

Commands Shipped

The full command surface, including the six run commands every run-producing team answers to.

Every call is myteams teamtester <command>. Installing the team also registers it as a Claude Code skill, so the same commands are reachable from inside a session as /myagenticteams:teamtester <command> — one definition, two front doors.

Commands

CommandWhat it does
myteams teamtester calibrateCheck every team's eval fixture; with --run, execute the measured calibration matrix and produce the capacity model.
myteams teamtester campaignStart, resume, or inspect a continuous test campaign (start --spec <file> | resume <id> | status <id>).
myteams teamtester analyzeMine a campaign's telemetry for systemic findings (--campaign <id>).
myteams teamtester campaign-reportWrite the human-readable campaign rollup (--campaign <id>).
myteams teamtester evidenceEmit one team's harness evidence bundle for the improve loop (--team <name>).
myteams teamtester report

The six it gets for free

Because this team declares a report command it produces runs, and every run-producing team answers the same six commands without declaring them. They default to the team's most recent run, so the common case takes no run id at all.

CommandWhat it does
myteams teamtester reportWrite the run's field report — the facts, plus the driver's own reflection on them.
myteams teamtester pauseStop the run at the next safe point, leaving it resumable.
myteams teamtester resumePick a paused run back up where it stopped.
myteams teamtester logThe run's event stream, as it happened.
myteams teamtester findingsEverything the run raised, filterable.
myteams teamtester insightsWhat the run learned that outlives it.