telemetry-summary

To track token usage and turns across multiple AI agents, you can aggregate per-task telemetry by consolidating turns, tokens, and delivered changes for each agent into a centralized, auditable ledger. This provides a single source of truth for resource consumption across your workflow.

An audit trail for AI subtask execution is maintained by invoking a telemetry aggregation step after each subtask completes, appending the latest per-agent and per-task telemetry to a central summary to ensure a complete, auditable record of resource consumption and delivered changes.

Task-level totals and Changes by Phase reporting are generated by reading subtask summaries and aggregating the per-agent telemetry data. This consolidated report is then produced for the chief-orchestrator to load into the task lifecycle for post-completion analysis.

Yes, telemetry aggregation can be used during audits or review cycles to verify workflow efficiency and identify bottlenecks. By tracking turns, token usage, and delivered changes per agent, you can pinpoint resource consumption issues across the pipeline.

Telemetry aggregation should be invoked after every subtask completes to append the latest telemetry to the central summary, and again at task completion. This ensures the final consolidated report contains a complete record of all agent activity and resource consumption.