🔴 Nightly eval regression — agent eval gate failing

[github-actions]

The nightly eval (eval-nightly.yml) failed — a canary failure or a >5% regression vs the recorded baseline.

• Run: https://github.com/InteractorOSS/product-manager/actions/runs/28276764575
• First red: 2026-06-27

Triage: open the run logs, find the failing agent/scenarios, and fix the prompt (src/server/agents/configs) or the scenario (evals/scenarios). This issue auto-closes when the nightly is green again.

[github-actions]

Still red on 2026-06-28. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28309655121

[github-actions]

Still red on 2026-06-29. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28346100400

[github-actions]

Still red on 2026-06-30. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28417512079

[github-actions]

Still red on 2026-07-01. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28490747582

[github-actions]

Still red on 2026-07-02. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28562470678

[github-actions]

Still red on 2026-07-03. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28641271277

[github-actions]

Still red on 2026-07-04. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28696589107

[github-actions]

Still red on 2026-07-05. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28731368043

[github-actions]

Still red on 2026-07-06. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28772176666

[github-actions]

Still red on 2026-07-07. Run: https://github.com/InteractorOSS/product-manager/actions/runs/28845243690