Emit structured workflow telemetry for AI agent runs. Use for run metrics, step latency, cost, tokens, model, loop count, failure rate, human-edit rate, escalations, council verdicts, validation status, PR outcomes, ML metrics, debug evidence, backprop data. Writes JSONL events to .workflow-runs/<ru
Emit structured workflow telemetry for AI agent runs. Use for run metrics, step latency, cost, tokens, model, loop count, failure rate, human-edit rate, escalations, council verdicts, validation status, PR outcomes, ML metrics, debug evidence, backprop data. Writes JSONL events to .workflow-runs/<run-id>/telemetry.jsonl or caller-provided store for later optimization.