Gpt 5 4 Prompting

This is internal plumbing for the Codex Claude Code plugin. It teaches Claude how to write tight, structured prompts for GPT-5.4 when Codex needs help with coding, debugging, or research tasks. The main idea is treating Codex like an operator, not a collaborator: use XML tags, define explicit output contracts, add verification loops for risky work, and skip the conversational padding. You get prompt recipes for common jobs like adversarial review or implementation tasks, plus a checklist to avoid vague nudges and unsupported guesses. Not user-invocable, so you won't call this directly. It's the meta-layer that keeps automated Codex calls crisp and grounded.

skills.sh installs: 738
GitHub stars: 35K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add openai/codex-plugin-cc --skill gpt-5-4-prompting --agent claude-code

Installs into .claude/skills of the current project.

Files

GPT-5.4 Prompting

Use this skill when codex:codex-rescue needs to ask Codex or another GPT-5.4-based workflow for help.

Prompt Codex like an operator, not a collaborator. Keep prompts compact and block-structured with XML tags. State the task, the output contract, the follow-through defaults, and the small set of extra constraints that matter.

Core rules:

Prefer one clear task per Codex run. Split unrelated asks into separate runs.
Tell Codex what done looks like. Do not assume it will infer the desired end state.
Add explicit grounding and verification rules for any task where unsupported guesses would hurt quality.
Prefer better prompt contracts over raising reasoning or adding long natural-language explanations.
Use XML tags consistently so the prompt has stable internal structure.

SKILL.md on GitHub