Skip to content
Skills Library
AI & agents · Review & audit

codex

/codex

Use when the user types /codex or asks for an external review, second opinion, sanity-check, or "ask codex" (RU «спроси codex», «внешнее мнение», «ревью от codex»). Also use proactively to pressure-test your own ideas, designs, plans, or diffs with an independent outside model before committing to them. Runs the codex CLI non-interactively in the background via `codex exec` and reports back its verdict.

Repository
quorum ↗
Licence
MIT
Last repository change
18.06.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 18.06.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Use when the user types /codex or asks for an external review, second opinion, sanity-check, or "ask codex" (RU «спроси codex», «внешнее мнение», «ревью от codex»). Also use proactively to pressure-test your own ideas, designs, plans, or diffs with an independent outside model before committing to them. Runs the codex CLI non-interactively in the background via `codex exec` and reports back its verdict.

Triggers from the description

ask codex
Original author description

Use when the user types /codex or asks for an external review, second opinion, sanity-check, or "ask codex" (RU «спроси codex», «внешнее мнение», «ревью от codex»). Also use proactively to pressure-test your own ideas, designs, plans, or diffs with an independent outside model before committing to them. Runs the codex CLI non-interactively in the background via `codex exec` and reports back its verdict.

How it works

  1. Prerequisites
  2. Default command
  3. Execution (background + wait)
  4. Prompt shape
  5. After it answers
Original SKILL.md headings · 47 lines, 456 words
SKILL.md preview
---
name: codex
description: Use when the user types /codex or asks for an external review, second opinion, sanity-check, or "ask codex" (RU «спроси codex», «внешнее мнение», «ревью от codex»). Also use proactively to pressure-test your own ideas, designs, plans, or diffs with an independent outside model before committing to them. Runs the codex CLI non-interactively in the background via `codex exec` and reports back its verdict.
---

# codex — external second opinion

Summon a fresh **Codex CLI** (OpenAI) as an independent outside model to challenge an idea, design, plan, or diff before acting on it. Use when the user writes `/codex "..."`, asks to involve codex, or whenever you want a second opinion that isn't anchored to your own reasoning.

## Prerequisites

- `codex` CLI installed and authenticated (`codex --version`).
- The default model is whatever the user's Codex config selects (e.g. `gpt-5.5`, `xhigh` reasoning) — strong for review. Do **not** override the model unless asked.

## Default command

Run from the relevant workspace root. Capture the clean final answer with `-o`; stdout also carries hook noise and a token-usage line, so never parse stdout for the answer:

```bash
codex exec -s read-only -o /tmp/codex-<tag>.md "<prompt>" > /tmp/codex-<tag>.log 2>&1
```

- Pick a unique `<tag>` per call (a short topic slug).
- `-s read-only` lets codex read the repo to ground its opinion but never modifies files — the right default for a review / second-opinion pass.
- `--skip-git-repo-check` if the workspace is not a git repo.
- `-C <dir>` to root codex somewhere other than the current directory.
- `-m <model>` to override the model — only on request.

## Execution (background + wait)

Codex with high reasoning takes minutes on a real review, so run it in the background and wait for it to finish — don't block the turn:

- **Claude Code:** launch via the Bash tool with `run_in_background: true`; you're re-invoked on exit. Then read `/tmp/codex-<tag>.md` for the verdict, and `/tmp/codex-<tag>.log` if it failed (non-zero exit or empty answer file).
- **Other agents / plain shell:** background it (`... &`) and wait for the process to exit, or run in the foreground with a generous timeout (codex is non-interactive and won't prompt).

Never treat streamed/partial stdout as the final answer — wait for exit and read the `-o` file.

## Prompt shape

Give codex enough to form an independent view:
- the workspace path and task context;
- the idea / design / plan / diff you want challenged, stated plainly;
- the exact question and focus areas (correctness, regressions, hidden assumptions, simpler alternatives, infra/production risk, tests);
- the output shape you want: verdict, where it agrees/disagrees, blocking vs non-blocking issues, concrete fixes.

Ask codex to **challenge** the idea and surface flaws and alternatives — the goal is an outside perspective, not validation. Default to read-only; use `-s workspace-write` only if the user explicitly wants codex to implement changes.

## After it answers

Reconcile codex's response against the actual code and facts before acting — it's an external opinion, not ground truth. Summarize the verdict and state plainly which points you agree with, which you reject, and why.

Versions

RepositoryScoreLicence
This page · alesha-pro/quorum100MIT
davila7/claude-code-templates100MIT
tigercosmos/japanese-podcast-learner100MIT
puck-muling/peer-agent100MIT
ispc/ispc100BSD-3-Clause
dzhng/skills100MIT
majiayu000/spellbook100MIT
onlypfachi/chief100MIT
dzhng/skills100MIT

Database snapshot: 13.11.2026 · package version d5de677569c0 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 95codex-exec
    Run OpenAI Codex CLI non-interactively to execute tasks, generate code, or get a second opinion. Use this skill whenever the user says "run codex", "ask codex", "codex exec", "have codex do", "send to codex", "pipe to codex", or wants to delegate a task to Codex for a second…
  • 100codex-review
    Use when you want an independent Codex review of a plan, spec or code change at a major checkpoint or before merging, when asked for a "codex review" or "second opinion", or before writing `codex exec` by hand.
  • 95codex-advisor
    This skill should be used when the user asks for a "GPT second opinion", wants a "cross-model review", needs to "check a plan before committing", wants another view after repeated failures, or explicitly invokes "codex-advisor".
  • 100codex-plan-review
    Get a second opinion on your current implementation plan by sending it to Codex for independent review. Use after brainstorming, after writing a plan, or when you want to iterate on a plan before execution. Triggers on 'get a second opinion', 'review plan with codex', 'codex…
  • 100review
    Cross-model orchestration between Claude Code and a local Codex CLI. Claude orchestrates, plans, and reviews; Codex implements and provides an independent second opinion. Use when you want two different model families to cross-check work rather than one model reviewing itself.
  • 100codex-review
    Get a second-opinion code review from OpenAI Codex via the `codex-run` CLI — dispatch it, wait on it correctly, verify what it found, and post it only when a human says so. Use ONLY when the user names Codex explicitly: "review this with codex", "codex review", "codex-review",…

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.