[github-actions]

Automated weekly refresh of patterns/ from active agentic workflow data.

[github-actions]

Reviewer: ponytail (data/content-only change; patterns JSON).

Not blocking, but worth a maintainer check before merging. The refresh looks like a partial scan rather than real drift:

• Sample sizes drop by about half or more: custom 469→221, code-improvement 86→40, dependency-monitor 62→22, documentation-updater 24→14, daily-test-improver 5→2, content-moderation 6→4.
• Some rates swing sharply on tiny samples. daily-test-improver goes 53%→0% (n=2) and its anti_patterns entry is cleared.
• The "recommended trigger" ordering changes in several archetypes, for example content-moderation now leads with schedule and custom with reaction.

The wizard and the .github/copilot-instructions text cite specific figures (e.g. n=86, 53% for daily-test-improver). If this merges, those figures become stale and trigger recommendations may shift because of a smaller sample. Please confirm the scan completed fully, for example by checking the data/ scan coverage or the workflow logs, and update any hard-coded stats and tests that reference them.

Generated by Specialist PR Review for #332 · copilot · auto · 14.5 AIC · ⌖ 0.498 AIC · ⊞ 7.9K