菜单
Send a prompt to Codex (OpenAI) and get a response. Use for code review, refactoring, architecture questions, or any task where a second model's perspective is valuable. Synthesize Codex output with your own analysis.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Log a gameplay decision, action, or event to the game log. Use at major decision points, after combat, after completing objectives, or when changing strategy.
Record a gameplay discovery or learning to the knowledge files. Use this whenever you discover something worth remembering — a new NPC, a quest clue, a useful item, a combat tactic, or world lore.
Search gameplay knowledge and history. Use when you need to remember something about a location, character, quest, item, or past decision.
基于 SOC 职业分类
| name | run-codex |
| description | Send a prompt to Codex (OpenAI) and get a response. Use for code review, refactoring, architecture questions, or any task where a second model's perspective is valuable. Synthesize Codex output with your own analysis. |
| allowed-tools | Bash, Read |
| argument-hint | <prompt> |
Send $ARGUMENTS as a prompt to Codex via codex exec. The entire argument string is the prompt — no mode parsing needed.
cd "$(git rev-parse --show-toplevel)"
OUTFILE=$(mktemp /tmp/codex_result.XXXXXX)
/opt/homebrew/bin/codex exec "$ARGUMENTS" \
-m gpt-5.3-codex \
-c model_reasoning_effort="high" \
--ephemeral --full-auto \
-o "$OUTFILE" 2>&1
Read $OUTFILE for the output. If the exit code is non-zero, read $OUTFILE for the error.