AgentSkillsCN

ask-codex

将任务委托给OpenAI Codex CLI进行审核，并获取反馈。当用户明确表示“让Codex来审阅”“请Codex帮忙审查”或“交由Codex处理”，或者希望从另一款AI那里获得关于代码或文本的第二意见，又或是希望通过Codex审阅命令对代码进行审查时，可使用此功能。

SKILL.md
--- frontmatter
name: ask-codex
description: >
  Delegate tasks to OpenAI Codex CLI for review and get responses back.
  Use when user explicitly asks to "ask codex", "let codex review", or "pass to codex",
  or wants a second opinion from another AI on code or text,
  or wants code review via codex review command.

Ask Codex

Invoke OpenAI Codex CLI to review content and return results for discussion.

Modes

1. Code Review Mode (codex review)

For reviewing git changes:

bash
# Review uncommitted changes
codex review --uncommitted

# Review against a branch
codex review --base main

# Review a specific commit
codex review --commit <sha>

# With custom instructions
codex review --uncommitted "Focus on security issues"

2. General Query Mode (codex exec)

For arbitrary prompts (read-only by default):

bash
# Ask codex a question (read-only sandbox)
codex exec -s read-only "Review this code and explain potential issues: <content>"

# Pass content via stdin
echo "content to review" | codex exec -s read-only -

Workflow

  1. •

    Determine appropriate mode based on user request:

    • •Git changes → use codex review
    • •Arbitrary content → use codex exec -s read-only
  2. •

    Execute codex command and capture output

  3. •

    Present codex's response to user

  4. •

    If issues found, discuss with user and offer to help resolve

Examples

User: "Ask codex to review my uncommitted changes"

bash
codex review --uncommitted

User: "Let codex review this function for bugs"

bash
codex exec -s read-only "Review this function for potential bugs and suggest improvements:

<paste function code here>"

User: "Get codex's opinion on my PR against main"

bash
codex review --base main

Important

  • •Default to read-only sandbox (-s read-only) unless user explicitly allows edits
  • •Always show codex's full response to user
  • •Discuss any issues or suggestions with user before taking action