Description
The fleet-wide TokenUsage=0 telemetry blackout has persisted ~13+ days: today's Detection Analysis (#43202) and API Consumption (#43181) both report all TokenUsage fields = 0 and empty token_usage.jsonl files, forcing token comparisons to be dropped across audits. The underlying bug is tracked (#42791), but nothing guards against regression.
Add a Smoke CI / self-check assertion that a representative agent run produces a non-empty token_usage.jsonl (and non-zero TokenUsage in run_summary.json). This fails fast the next time the emitter breaks, instead of silently blinding every downstream audit for weeks.
Expected Impact
Catches token-telemetry regressions in hours rather than weeks; protects api-consumption, detection, and cost audits.
Suggested Agent
test-quality or compiler-quality agent.
Estimated Effort
Medium (1-4 hours)
Data Source
DeepReport Intelligence analysis run 2026-07-03 (workflow-run 28669423540); #43202, #43181.
Generated by 🔬 Deep Report · 171.9 AIC · ⌖ 17.4 AIC · ⊞ 10.2K · ◷
Description
The fleet-wide
TokenUsage=0telemetry blackout has persisted ~13+ days: today's Detection Analysis (#43202) and API Consumption (#43181) both report allTokenUsagefields = 0 and emptytoken_usage.jsonlfiles, forcing token comparisons to be dropped across audits. The underlying bug is tracked (#42791), but nothing guards against regression.Add a Smoke CI / self-check assertion that a representative agent run produces a non-empty
token_usage.jsonl(and non-zeroTokenUsageinrun_summary.json). This fails fast the next time the emitter breaks, instead of silently blinding every downstream audit for weeks.Expected Impact
Catches token-telemetry regressions in hours rather than weeks; protects api-consumption, detection, and cost audits.
Suggested Agent
test-quality or compiler-quality agent.
Estimated Effort
Medium (1-4 hours)
Data Source
DeepReport Intelligence analysis run 2026-07-03 (workflow-run 28669423540); #43202, #43181.