[copilot-opt] Reduce 6-reviewer fan-out per Copilot PR push to cut action_required noise by 50%+

[github-actions]

Problem

Every Copilot agent PR push triggers 6 review workflows simultaneously (Q, Grumpy Code Reviewer 🔥, Scout, PR Nitpick Reviewer 🔍, Security Review Agent 🔒, /cloclo). All 6 complete with action_required within seconds of each push, generating 6 separate human-attention items and 6 independent workflow runs for every single commit. This fan-out saturates the review queue and obscures actionable signal under noise.

Evidence

• Analysis window: 2026-04-19 (today; only date present in session data)
• Sessions analyzed: 50 workflow runs across 4 Copilot branches
• Key metrics and examples:
  • 36 of 50 session runs (72 %) concluded with action_required — all from review workflows
  • For each push to copilot/optimize-token-usage-optimizer (00:50:36 Z): Q, Grumpy Code Reviewer, Scout, PR Nitpick Reviewer, Security Review Agent, and /cloclo all fired and completed with action_required within the same second
  • For each push to copilot/analyze-copilot-sessions (00:34:52 Z): all 6 review workflows again fired simultaneously — contributing 24 of the 33 total runs on that branch alone
  • Only 11 of the 50 runs (22 %) were substantive CI jobs (Smoke CI, CGO, Design Decision Gate, Test Quality Sentinel, Doc Build, Content Moderation)

Proposed Change

Consolidate or gate the parallel reviewer workflows to reduce fan-out:

1. Short-term: Add a concurrency group to each review workflow so that only one instance runs per PR at a time, cancelling in-progress runs on new pushes.
2. Medium-term: Merge the style/nitpick reviewers (Grumpy Code Reviewer, PR Nitpick Reviewer, /cloclo) into a single "Style & Quality" composite workflow that outputs one consolidated comment.
3. Long-term: Gate non-critical reviewers (Scout, Grumpy Code Reviewer, PR Nitpick Reviewer, /cloclo) behind the critical path (Security Review Agent, Q) so they only run after the first two pass, reducing unnecessary fan-out on broken PRs.

Expected Impact

• Reduce review workflow runs per PR push from 6 to 2–3
• Cut action_required queue noise by ~50–67 %
• Reduce total CI-minutes spent on review fan-out by an estimated 60 % (6 independent runs → 2)
• Make actionable security and quality signals easier to triage by reducing noise

Notes

• Distinct root cause category: review workflow fan-out / over-triggering
• Data quality caveats: session events.jsonl unavailable; latency data per reviewer is absent; fan-out count is based on workflow run metadata only
• PR data corroboration: 204 / 1,000 closed-without-merge PRs suggests high review friction is a contributing factor to abandonment

References:

• §24617578848 — Q (action_required) on copilot/optimize-token-usage-optimizer
• §24617339322 — Security Review Agent (action_required) on copilot/analyze-copilot-sessions
• §24617578805 — /cloclo (action_required) on copilot/optimize-token-usage-optimizer

Generated by Copilot Opt · ● 2.1M · ◷

[github-actions]

🍪 *Issue Monster selected this for Copilotecho BEGIN___COMMAND_OUTPUT_MARKER ; PS1= ; PS2= ; unset HISTFILE ; EC=0 ; echo ___BEGIN___COMMAND_DONE_MARKER___0 ; }

I've identified this issue as a good candidate for automated resolution and requested assignment to the Copilot coding agent.

If assignment succeeds, the Copilot coding agent will analyze the issue and create a pull request with the fix.

Om nom nom! 🍪

🍪 Om nom nom by Issue Monster · ● 11.5K · ◷