Description
Three skills/quality reviewers — Matt Pocock Skills Reviewer, Impeccable Skills Reviewer, and Design Decision Gate — run on essentially every PR and together consume ~180 action-minutes in a single 40-run window (the largest cost center in the fleet, ~12+ runs/window). Most PRs do not touch the skills/design surface these reviewers target.
Fix
Add pull_request.paths filters (or an early skip-if-no-relevant-change guard) so each reviewer only triggers when files in its domain change (e.g. .github/skills/**, design/spec dirs). Consider consolidating overlapping reviewers.
Expected Impact
Cuts a large recurring action-minute + token cost center while preserving coverage on the PRs that actually matter.
Suggested Agent
Workflow Health Manager / Copilot SWE Agent.
Estimated Effort
Medium (1-4 hours)
Data Source
DeepReport 2026-06-29; Weekly Workflow Analysis #42229.
Generated by 🔬 DeepReport - Intelligence Gathering Agent · 212.1 AIC · ⌖ 11.9 AIC · ⊞ 9.8K · ◷
Description
Three skills/quality reviewers — Matt Pocock Skills Reviewer, Impeccable Skills Reviewer, and Design Decision Gate — run on essentially every PR and together consume ~180 action-minutes in a single 40-run window (the largest cost center in the fleet, ~12+ runs/window). Most PRs do not touch the skills/design surface these reviewers target.
Fix
Add
pull_request.pathsfilters (or an early skip-if-no-relevant-change guard) so each reviewer only triggers when files in its domain change (e.g..github/skills/**, design/spec dirs). Consider consolidating overlapping reviewers.Expected Impact
Cuts a large recurring action-minute + token cost center while preserving coverage on the PRs that actually matter.
Suggested Agent
Workflow Health Manager / Copilot SWE Agent.
Estimated Effort
Medium (1-4 hours)
Data Source
DeepReport 2026-06-29; Weekly Workflow Analysis #42229.