feat(ci): disclose AI-generated CLI diagnoses

[121watts]

Summary

Adds a tiny honesty footer to CLI failure diagnoses, once and out of the way.

What was happening

The human-readable CLI output presented generated diagnoses and possible fixes without indicating that the guidance came from AI or could be wrong.

What happens now

Grouped and focused diagnosis output ends with “This diagnosis is AI-generated and can make mistakes.” when any diagnosis or possible fix was rendered. The footer appears once for the whole output, never once per failure, and stays absent when there is no AI guidance.

JSON output is unchanged.

Anything else

This complements the app disclosure PR and the Opus 4.7 API upgrade.

---

Note

Low Risk
CLI presentation-only change with no API or JSON behavior changes.

Overview
Human-readable depot ci diagnose output now ends with a single footer when any Diagnosis or Possible fix text was shown: "This diagnosis is AI-generated and can make mistakes."

The footer is printed once per command (after bounds/summary notes), including when multiple failure groups or representative attempts have AI guidance. It is omitted when neither field is present. JSON output is unchanged.

Tests cover grouped and focused paths, multi-group/attempt cases, diagnosis-only vs possible-fix-only, and the no-AI case.

^{Reviewed by Cursor Bugbot for commit fdcfd47. Bugbot is set up for automated code reviews on this repo. Configure here.}

[linear-code]

DEP-4954

[cursor]

Approved: Cursor Bugbot and Cursor Security Agent both completed successfully with no findings requiring human review. No reviewers were assigned.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}