[0.146] Backport safer cyber-model auto-review defaults - #37057
Conversation
- 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. - Cover auto-review defaults, requirement-aware fallback behavior, permission preservation across reasoning changes, and the new warning notice. GitOrigin-RevId: 8f25b0082d3051d9ff91f7355921236d0f82f758
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 442ae3a7f0
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| &workspace_profile, | ||
| ); | ||
|
|
||
| if workspace_allowed && let Some(reviewer) = cyber_model_approval_reviewer(&self.config) |
There was a problem hiding this comment.
When requirements disallow selecting :workspace but still allow on-request and the auto-review/user reviewer, this workspace_allowed && guard prevents sending any cyber-model safety defaults at all. In that managed-profile setup a user switching from a previous unsafe profile (for example full access with never) remains on the old approval policy/reviewer, even though the approval defaults are independently representable in ThreadSettingsUpdateParams; split the permission-profile override from the approval/reviewer overrides so the available safeguards still apply.
Useful? React with 👍 / 👎.
2 participants
Backport cyber-model auto-review defaults to the 0.146 stable release line.
Cherry-pick #37055. Omit precomputed schema artifacts that do not exist on release/0.146.
Validation: git diff --check.