Prompting operators' guide for Codex and GPT-5.4: XML-structured contracts and recipes
Block-structured prompt recipe (task, output contracts, follow-through policy, verification and grounding loops), reusable prompt blocks, end-to-end Codex recipes, and antipatterns — internal plugin skill (user-invocable: false)
- What
- Block-structured prompt recipe (task, output contracts, follow-through policy, verification and grounding loops), reusable prompt blocks, end-to-end Codex recipes, and antipatterns — internal plugin skill (user-invocable: false)
- Cost
- Free
- Needs
- the Codex Claude Code plugin (codex-plugin-cc) — this is an internal (user-invocable: false) skill that activates via codex:codex-rescue; installing it standalone only makes sense as reference material
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — @openai's gpt-5-4-prompting skill from the Codex Claude Code plugin: the operator's recipe for composing Codex/GPT-5.4 prompts — prompt like an operator, not a collaborator: one clear task per run, compact block-structured prompts with stable XML tags (`<task>`, `<structured_output_contract>`, `<default_follow_through_policy>`, `<verification_loop>`/`<completeness_contract>`, `<grounding_rules>`/`<citation_rules>`), explicit grounding and verification rules wherever unsupported guesses would hurt, better prompt contracts before raising reasoning, and a prompt-assembly checklist (define task, smallest output contract, follow-through decision, verification/grounding/safety tags where needed, strip redundancies). Covers block selection per job type (coding/debugging → verification + missing-context gating; review → grounding + dig-deeper nudge; research → research mode + citation rules; write-capable → action safety), command choice (`review`/`adversarial-review` for local git diffs, `task` for diagnosis/planning/research/implementation, `--resume-last` for deltas), and reusable blocks plus concrete recipes in `references/`. Honest caveats: this is an INTERNAL plugin skill (`user-invocable: false`) — it activates via `codex:codex-rescue`, not by direct user invocation; the referenced blocks and recipes ship with the plugin, so install it as part of the plugin bundle for full value. Apache-2.0 licensed. Skill Harbor never reviews the code, review it yourself before use. Discovered via skills.sh.
Version:
Install
Prerequisites: the Codex Claude Code plugin (codex-plugin-cc) — this is an internal (user-invocable: false) skill that activates via codex:codex-rescue; installing it standalone only makes sense as reference material Install "Prompting operators' guide for Codex and GPT-5.4: XML-structured contracts and recipes" for me. It gives my agent @openai's Codex prompting playbook: block-structured XML prompt recipes (task, output contracts, follow-through policy, verification and grounding loops), block selection per job type (coding, review, research, write-capable), command choice (review/adversarial-review/task/--resume-last), a prompt-assembly checklist, and reusable prompt blocks, recipes, and antipatterns from references/. Apache-2.0 licensed. Repository: https://github.com/openai/codex-plugin-cc/blob/main/plugins/codex/skills/gpt-5-4-prompting/SKILL.md 1. Fetch the SKILL.md file (and any helper files) from the repository path into a temporary folder and summarize what it does in one or two sentences. 2. Safety check: review the SKILL.md and scripts for anything suspicious (unexpected network calls, shell commands, credential harvesting). This repo should contain zero secrets in code, credentials only via the secure vault, allowed hosts declared in the SKILL.md. Verify that holds here; STOP on any red flag and tell me. 3. Install it as a skill: copy SKILL.md and its helper files into the agent's skills directory, in a folder named "gpt-5-4-prompting". 4. Verify with no network calls: frontmatter valid, files in place. 5. Report what was installed, where, and what I still need to do myself (e.g. keep it inside the Codex plugin bundle — it is internal (user-invocable: false) and activates via codex:codex-rescue, not by direct user invocation; the reusable prompt blocks and recipes it references ship with the plugin). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.
Questions
How do I install a build?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the build up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a build before installing it.