From codex
Codex-powered plan review, completion verification, and code review. This skill should be used on multi-file changes or after context compactions.
How this skill is triggered — by the user, by Claude, or both
Slash command
/codex:reviewThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Consult the [`codex` skill](../codex/SKILL.md) for MCP tool usage, thread configuration, and thread lifecycle.
Consult the codex skill for MCP tool usage, thread configuration, and thread lifecycle.
Use Codex for:
Skip for: simple docs, single-file changes, trivial fixes.
All workflows follow this loop:
codex tool (first call) or codex-reply tool (follow-ups)Expect multi-minute responses; don't cancel early. Not all feedback must be incorporated - push back on overly pedantic points with reasoning.
Use before exiting plan mode.
Context: [user's request and goals]
Scope: [in/out of scope]
Constraints: [requirements]
Plan: [path or inline content]
Review for completeness and correctness. Respond APPROVED if solid.
Use before telling the user work is complete, especially after:
Implementation complete. Examine files on disk and verify:
- All planned changes were made
- No partial implementations remain
- No steps discussed but not executed
List anything incomplete.
If incomplete, fix and re-verify.
Use after writing significant code.
Purpose: [what the code should do]
Files: [paths or inline content]
Concerns: [specific areas to check]
Test results: [if available]
Review for correctness and issues.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub tvishwanadha/skills --plugin codex