Patch release refining reasoning effort flag aliases, permission mode commands, and related documentation and shell completions.
Changelog
[changed] --reasoning-effort and --effort are now documented as interchangeable aliases for setting reasoning effort levels; both work in the TUI and headless mode.
[changed] Updated /always-approve documentation to cover the new /auto command and clarified how the two act as toggles in the completion menu depending on the current permission mode.
[changed] Added auto_interject_on_task_wait configuration option (defaults to true) that sends queued follow-ups immediately when the agent is blocked on task output.
[changed] Shell completions for bash, zsh, fish, elvish, and PowerShell updated to reflect the unified effort flags and removed the old discrete effort enum from --effort.
[changed] Headless mode documentation updated to note that --reasoning-effort/--effort and --permission-mode now apply in both modes, while keeping the list of headless-only flags accurate.
Under the hood
Internal version metadata and bundled manifest checksums refreshed for 0.2.89.
Minor runtime path references updated in the help skill.
