aar

You can automate after-action reviews by invoking the skill with a defined scope such as a time window or a single session. It aggregates data from the orchestrator DB, A2A endpoints, pm2 logs, and git history to produce a structured AAR document.

Gathering forensics for a postmortem is done by applying a defined scope to aggregate data across the orchestrator DB, A2A endpoints, pm2 logs, git history, and queue files. This captures the evidence needed to classify failures into predefined modes.

Failure-mode classification for agent dispatch postmortems categorizes aggregated forensics data from your orchestrator DB and logs into predefined categories. This classification guides remediation by surfacing actionable insights within the structured AAR document.

Yes, you can run an after-action review on a single agent session. The skill supports applying its forensics gathering and failure classification to either a defined time window or a single specific session scope.

You need access to the orchestrator DB, A2A endpoints, pm2 logs, git history, and queue files for an agent run postmortem. The skill aggregates data from these sources to build an evidence-based narrative and capture benchmarks.

Structured after-action reviews provide a mechanism to capture and learn from orchestrated agent runs by surfacing actionable insights from forensics data. This replaces ad hoc recollections with an evidence-based narrative and benchmarks stored for knowledge sharing.