Description
The [api-consumption] report (discussion #46226) found run_summary.json's authoritative SafeItemsCount field is unpopulated (returns 0) across every run in the current CLI version — the same gap that forced the 07-15 and 07-16 history entries to 0. Audits now fall back to summary.json -> safe_outputs.total_items. Fix the run-summary writer to populate SafeItemsCount from actuated safe-output items so downstream audits report accurate write counts directly.
Expected Impact
Restores accurate safe-output write metrics across multiple daily audits (API Consumption, Safe Output Health) without brittle fallbacks; improves historical trend fidelity.
Suggested Agent
Copilot SWE Agent (Go, gh-aw logs/reporting internals).
Estimated Effort
Medium (1-4 hours).
Data Source
DeepReport 2026-07-17 - [api-consumption] discussion #46226. Distinct from the closed token-total gap #45511; dedup-checked: no open issue covers SafeItemsCount.
Generated by 🔬 Deep Report · 208.7 AIC · ⌖ 11.4 AIC · ⊞ 10K · ◷
Description
The [api-consumption] report (discussion #46226) found run_summary.json's authoritative SafeItemsCount field is unpopulated (returns 0) across every run in the current CLI version — the same gap that forced the 07-15 and 07-16 history entries to 0. Audits now fall back to summary.json -> safe_outputs.total_items. Fix the run-summary writer to populate SafeItemsCount from actuated safe-output items so downstream audits report accurate write counts directly.
Expected Impact
Restores accurate safe-output write metrics across multiple daily audits (API Consumption, Safe Output Health) without brittle fallbacks; improves historical trend fidelity.
Suggested Agent
Copilot SWE Agent (Go, gh-aw logs/reporting internals).
Estimated Effort
Medium (1-4 hours).
Data Source
DeepReport 2026-07-17 - [api-consumption] discussion #46226. Distinct from the closed token-total gap #45511; dedup-checked: no open issue covers SafeItemsCount.