AgentSkillsCN

claude-parity

针对 Codex CLI,提供与 Claude Code 一致的命令行工作流。当用户希望以斜杠命令的形式完成上下文引导、提交代码、创建 Pull Request、审查 Pull Request、调试程序、执行测试、管理依赖项、进行安全审计、制定功能规格,或构建流水线时,可使用此技能。

SKILL.md
--- frontmatter
name: claude-parity
description: Claude Code parity workflows for Codex CLI. Use when users ask for slash-command style tasks like context-prime, commit, pr-create, review-pr, debug, test, dependencies, security-audit, feature-spec, or pipeline.
metadata:
  short-description: Claude-style workflow playbooks

Claude Parity

This skill mirrors the existing Claude Code command workflow style inside Codex.

When to Use

Use this skill when:

  • The user explicitly references Claude Code workflows or slash commands.
  • The user asks for one-shot operational tasks (commit, pr-create, review-pr, security-audit, etc.).
  • The user wants staged delivery flow (pipeline).

How to Run This Skill

  1. Map user intent to a command flow using references/command-mapping.md.
  2. Execute the flow directly (do not ask for unnecessary confirmation).
  3. Keep output practical: findings first, then actions taken, then next steps.
  4. Apply guardrails from ~/.codex/AGENTS.md or repo AGENTS.md.

Pipeline Mode

For feature pipeline requests, follow references/pipeline.md exactly and pause between stages.

Output Standards

  • review output: severity-ordered findings with concrete file references.
  • commit output: final message before commit action.
  • pr-create output: title, summary bullets, testing notes, breaking changes, linked issues.
  • debug output: evidence, root cause, fix, regression test.