[automated] Monitor selective PR CI accuracy after enforcement

[radical]

[automated] Track the bounded rollout audit after #19688 enabled selective PR test execution.

Current checkpoint: 42 of 50 eligible runs reviewed; 30 selected a subset and 12 selected ALL. No false positives or false negatives have been observed. One safe over-selection issue in renamed-file handling was investigated.

• Continue through September 4, 2026 or 50 eligible runs.
• Record missed test lanes, unnecessary selections, and incorrect ALL fallbacks.
• File and land fixes for confirmed selector defects.
• Publish the final accuracy and performance summary in Tracking: CI improvements #18036.

[joperezr]

🤖 Aspire triage pass — 2026-08-28 22:33 UTC

Area: area-engineering-systems
Kind: feature-request (internal CI rollout audit / monitoring task)
Repro: none — monitoring/audit tracking issue for the rollout enabled by #19688
Affected version(s): N/A — internal CI infrastructure (not a shipped product surface)
Severity: S4 — monitoring/accuracy audit of selective PR test execution; internal tooling

Urgency: ⚪ backlog — no milestone
Reason: bounded internal rollout audit tracked against #19757; not a regression or customer-facing bug

Suggested owner(s): David Pine (@IEvangelist) (score 24), backup Adam Ratzman (@adamint) (score 12)
Evidence:

• IEvangelist merged Fix docs safe-output base resolution #19446, Fix gh-aw v0.86.2 safe-output base parsing #19445, Upgrade gh-aw compiler to v0.86.2 #19438, Fix PR docs recovery base handling #19236 in area-engineering-systems (last 90d, 8 PRs)
• adamint merged Remove pipeline-scoped Publish-Build-Assets group #19518, Avoid MongoDB credential false positive #19305, Re-enable VS Code extension E2E tests in CI #19142 in area-engineering-systems (last 90d, 4 PRs)
• (issue author excluded from owner scoring; no file paths cited so CODEOWNERS not scored)

Confidence: medium
Needs human?: no

[radical]

[automated] Follow-up found during continued monitoring: #19986 tracks benchmark projects outside the Aspire.slnx graph that are currently reported as unattributed and conservatively fall back to ALL.

Run 34247134578 exposed the gap without losing coverage because Directory.Packages.props independently selected ALL.