Default Agent Sessions to off#2039
Conversation
![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: 293f26589b
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".
| forKey: "providersSortedAlphabetically") as? Bool ?? false | ||
| let appLanguageRaw = userDefaults.string(forKey: "appLanguage") | ||
| let agentSessionsEnabled = userDefaults.object(forKey: "agentSessionsEnabled") as? Bool ?? true | ||
| let agentSessionsEnabled = userDefaults.object(forKey: "agentSessionsEnabled") as? Bool ?? false |
There was a problem hiding this comment.
For existing installs that saved agentSessionsManualHosts while the old default kept Agent Sessions enabled, the agentSessionsEnabled key can still be absent because editing the host field does not write this Boolean. After this change, that same state loads as disabled, and AgentSessionsStore ignores the configured hosts until the user rediscovers and re-enables the toggle, so manual remote session discovery silently regresses for those users. Consider treating non-empty manual hosts as an opt-in during this default migration or writing a one-time migrated value.
Useful? React with 👍 / 👎.
1 participant
Summary
Verification
make checkmake test(598 selections, 50 groups, 0 failures or retries)swift test --no-parallel --filter 'SettingsStoreCoverageTests|AgentSessionMenuDescriptorTests'after rebasing onto latestmainagentSessionsEnabled = false; bundledsessions --helpremains availableUI accessibility automation could launch CodexBar and open its settings/menu, but those controls did not expose usable accessibility elements, so visual menu extraction was inconclusive. The headless menu-descriptor regression test covers section suppression and explicit enablement.