Menu
Use when the user asks to run Kimi CLI in non-interactive mode or references Kimi for AI-assisted coding or agent tasks.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | kimi-cli |
| description | Use when the user asks to run Kimi CLI in non-interactive mode or references Kimi for AI-assisted coding or agent tasks. |
Before running Kimi, follow the shared logging and summary protocol in ../_shared/agent-cli.md.
command -v kimi--print mode for non-interactive use. For implementation tasks, first use superpowers:using-git-worktrees to create a fresh git worktree, then run Kimi there with write-capable/auto-approve options when approved.the current host's user-question or approval mechanism before using -y, --yolo, --yes or --print modes (they auto-approve actions).-p, --prompt, --command <text> - User prompt to the agent (required for non-interactive)-C, --continue - Continue previous session for working directory-S, --session <id> - Session ID to resume-m, --model <model> - LLM model to use-y, --yolo, --yes - Automatically approve all actions (use with caution)--print - Run in print mode (non-interactive, implies --yolo)--quiet - Alias for --print with text output and final message only-w, --work-dir <directory> - Working directory for the agent--add-dir <directory> - Add additional directory to workspace scope--agent <default|okabe> - Builtin agent specificationfull.md, and require the prompt to write summary.md.| Use case | Command pattern |
|---|---|
| Non-interactive print mode | kimi --print --prompt "prompt" |
| Quiet mode (final msg only) | kimi --quiet --prompt "prompt" |
| Continue last session | kimi --print --continue --prompt "follow-up" |
| Resume specific session | kimi --print --session <id> --prompt "prompt" |
| Yolo mode in fresh worktree only (auto-approve) | kimi --print --yolo --prompt "prompt" |
| With specific model | kimi --print --model <model> --prompt "prompt" |
| Additional directory | kimi --print --add-dir /path --prompt "prompt" |
| Okabe agent | kimi --print --agent okabe --prompt "prompt" |
| Mode | Description | Use Case |
|---|---|---|
--print | Non-interactive, auto-approves | Analysis only unless running inside a fresh worktree |
--quiet | Non-interactive, only final output | Clean output for scripts |
--yolo | Auto-approve all actions | Full access in a fresh worktree |
# Non-interactive print mode; analysis in main tree, writes only in a fresh worktree
kimi --print --prompt "Refactor the utils module"
# Quiet mode (only final message)
kimi --quiet --prompt "Fix the bug in src/main.py"
# Continue last session
kimi --print --continue --prompt "What else can you improve?"
# With specific model and okabe agent
kimi --print --agent okabe --model moonshot-v1-auto --prompt "Analyze codebase"
# Capture stdout/stderr according to ../_shared/agent-cli.md
kimi command, immediately use the current host's user-question or approval mechanism to confirm next steps, collect clarifications, or decide whether to resume with --continue.kimi --version or a kimi command exits non-zero; request direction before retrying.--yolo, --yes, --print) ask the user for permission using the current host's user-question or approval mechanism unless it was already given.the current host's user-question or approval mechanism.Use when installing, reinstalling, auditing, or preparing Brain-managed Superpowers, Spec Kit, and OpenSpec workflow arsenals under a project-local `.brain/vendor` directory so Brain can inspect and route them without exposing their raw Codex skills/prompts at the project root.
Use when the user asks to check, add, or manage macOS Calendar events.
Use when the user asks to send an email via the macOS Mail app.
Use when the user asks to send a text message or iMessage via the macOS Messages app.
Use when a task involves web search, current information, source verification, citations, or entity/person/paper/project disambiguation.
Use when the task involves `exec_command` escalation, `prefix_rule`, sandbox permissions, writable roots, or Codex platform-specific operational patterns.