Add safer TUI defaults for cyber models (#37055)

## What changed

- Propagate the optional `modelSpecialty` field through the model catalog and
  app-server `model/list` response.
- When a newly selected model has the `cyber` specialty, default an active TUI
  thread to workspace-write permissions with on-request approval. Use auto
  review when available, otherwise keep the user as reviewer, while respecting
  configured permission requirements.
- Show a notice when auto review is applied and strengthen the full-access
  warning for cyber models.
- Preserve explicitly selected permissions when only reasoning settings change.

## Testing

- Cover auto-review defaults, requirement-aware fallback behavior, permission
  preservation across reasoning changes, and the new warning notice.

GitOrigin-RevId: 8f25b0082d3051d9ff91f7355921236d0f82f758
This commit is contained in:
Won Park authored and copyberry committed 2026-08-05 08:04:15 +00:00
1 parent 3171166881
commit 757c151a0e
35 files changed
+536 -39

No files matched your search

@@ -3141,6 +3141,7 @@ async fn model_picker_hides_show_in_picker_false_models_from_cache() {
model: slug.to_string(),
display_name: slug.to_string(),
description: format!("{slug} description"),
model_specialty: None,
default_reasoning_effort: ReasoningEffortConfig::Medium,
supported_reasoning_efforts: vec![ReasoningEffortPreset {
effort: ReasoningEffortConfig::Medium,
@@ -3635,6 +3636,7 @@ async fn single_reasoning_option_skips_selection() {
model: "model-with-single-reasoning".to_string(),
display_name: "model-with-single-reasoning".to_string(),
description: "".to_string(),
model_specialty: None,
default_reasoning_effort: ReasoningEffortConfig::High,
supported_reasoning_efforts: single_effort,
supports_personality: false,