This was generated by AI during triage.
Problem
Desktop Background agents list and detail views do not show which model performed the delegated work.
Task names sometimes mention a model, but they are user-authored and are not reliable metadata. Runtime model metadata is already persisted per session and usage event.
Motivation
Model visibility helps users audit delegated work, compare agent runs, and understand Auto routing or differences in output.
Proposed solution
Enrich the existing Desktop surfaces:
- Background agents list: show a short effective-model badge.
- Agent detail/header: show the model display name and stable model ID.
- Support truncation and tooltips; do not rely on color alone.
Display semantics
- Running: latest effective model.
- Completed: effective model used.
- Auto: concrete routed model, with configured
Auto optionally shown secondarily.
- Unavailable:
Unknown; never infer from the parent session.
- Multiple models:
Mixed in the list, with models shown per turn or as a list in details.
Terminology should align with the effective-model states introduced by #3824.
Example workflow
- Delegate several background agents using pinned and Auto-selected models.
- Open Background agents to compare the effective models.
- Open an agent to see its full model identity.
Acceptance criteria
- List rows show the effective model when available.
- Detail headers show the display name and stable model ID.
- Auto-routed agents show the concrete effective model.
- Missing metadata displays
Unknown.
- Multi-model agents display
Mixed, with details available in the agent view.
- A live
Resolving... transition is optional.
Out of scope
- A new CLI/TUI worker panel.
- Reasoning-effort or context-tier display.
- Changing model selection or routing.
- Making the subagent aware of its model.
Related issues
Environment
- GitHub Copilot desktop app v1.0.25
- App-spawned CLI v1.0.71
- Windows 11
Cropped screenshots can be attached showing the missing model field in the Background agents list and agent header.
Problem
Desktop Background agents list and detail views do not show which model performed the delegated work.
Task names sometimes mention a model, but they are user-authored and are not reliable metadata. Runtime model metadata is already persisted per session and usage event.
Motivation
Model visibility helps users audit delegated work, compare agent runs, and understand Auto routing or differences in output.
Proposed solution
Enrich the existing Desktop surfaces:
Display semantics
Autooptionally shown secondarily.Unknown; never infer from the parent session.Mixedin the list, with models shown per turn or as a list in details.Terminology should align with the effective-model states introduced by #3824.
Example workflow
Acceptance criteria
Unknown.Mixed, with details available in the agent view.Resolving...transition is optional.Out of scope
Related issues
Environment
Cropped screenshots can be attached showing the missing model field in the Background agents list and agent header.