fix(cli): clarify permission mode approval behavior

[whatevertogo]

Related Issue

Closes #1866
Closes #1874
Addresses #1859 — clarifies the semantics without renaming the shipped modes.

Problem

The permission engines implement an increasing autonomy order of manual < yolo < auto:

• Manual reviews actions that require approval.
• YOLO auto-approves regular tools, while sensitive actions may still need approval and the agent may still ask questions.
• Auto is the unattended mode: permission prompts are approved automatically and the agent does not ask questions.

The copy shipped across the CLI and TUI used absolute or incomplete descriptions such as "Approve every action yourself", "Tool actions auto-approved", and "All actions auto-approved". That made the distinction unclear and contributed to the confusion reported in #1859, #1866, and #1874.

Issue #1859 suggests swapping or renaming the modes. This PR takes the compatibility-preserving approach instead: it keeps the shipped names, command semantics, persisted values, and permission-engine behavior unchanged, while making the user-facing contract explicit.

This is a follow-up to #1867, which first aligned the broad permission-mode descriptions.

What changed

• Clarified the permission descriptions shown by CLI help, the /permission selector, slash-command autocomplete, /auto and /yolo notices, and session replay.
• Kept the selector ordered as Manual → YOLO → Auto to match increasing autonomy.
• Added a patch changeset for @moonshot-ai/kimi-code, as required for this user-visible CLI/TUI change.
• Preserved the existing local source layout instead of applying unrelated whole-file formatting.
• Did not add copy-specific assertions: the existing behavior-level tests already cover these surfaces, and duplicating static strings would make the suite brittle.

File change summary

• Added: 1 file
• Modified: 5 files
• Deleted: 0 files

Area	Files	Purpose
Release	.changeset/clarify-permission-mode-copy.md	Record the user-visible patch
CLI/TUI copy	apps/kimi-code/src/cli/commands.ts, src/tui/commands/config.ts, src/tui/commands/registry.ts	Align help, notices, and command descriptions
Selector/replay	src/tui/components/dialogs/permission-selector.ts, src/tui/controllers/session-replay.ts	Clarify visible mode behavior

Verification

• Existing targeted Vitest suite: 4 files, 199 tests passed.
• apps/kimi-code TypeScript typecheck passed.
• Oxlint passed with 0 warnings and 0 errors on the changed source files.
• git diff --check origin/main...HEAD passed.
• Read-only public-text audit found no context-specific internal identifiers or agent attribution.

Checklist

• I have read the CONTRIBUTING document.
• I have linked a related issue, or explained the problem above.
• I have added tests that prove my feature works. Not applicable: this is a copy-only change, and the existing behavior-level tests pass.
• Ran gen-changesets skill, or this PR needs no changeset.
• Ran gen-docs skill, or this PR needs no doc update. The current docs already describe the engine semantics correctly; this PR only makes existing CLI/TUI copy more precise.

[changeset-bot]

🦋 Changeset detected

Latest commit: 0b56643

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@moonshot-ai/kimi-code	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR