You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[observability escalation] Smoke Claude and Smoke Copilot repeatedly exceed resource and control thresholds #27637
Two workflows crossed the manual escalation thresholds over the 14-day window. Both are smoke/validation workflows behaving like frontier-model research agents — a structural mismatch that drives avoidable cost and poor control.
Workflows Requiring Owner Attention
1. Smoke Copilot — 🔴 Highest priority
Threshold crossed: 3/3 runs with resource_heavy_for_domain HIGH; 3/3 runs with poor_agentic_control MEDIUM
Evidence: turns=15–20, write_actions=28–30 per run, duration 10–14 min, execution=exploratory, actuation=write_heavy
Recommended action: Add explicit turn limits (max_turns) and tighten instructions to prevent exploratory write-heavy behavior in a smoke context. A smoke test should be directed + read_only or minimal-write.
Recommended action: Move data-fetching turns to deterministic pre-steps (frontmatter steps). Reduce tool breadth. A smoke test does not need 26 tool types.
Activity