agentsclimarketplace
Sign in

Consult codex

Skill adeilsonrbrito/cross-model-consult/consult-codex

Consult the Codex CLI for an independent, cross-model second opinion on a decision, design, interpretation, diagnosis, or piece of work. Codex runs a different model and provider, so its take is genuine divergence — not an echo of your own reasoning, which is what makes it useful. Reach for it on a genuinely hard or high-stakes call: a decision that is costly or hard to reverse, a design or plan worth pressure-testing before committing to it, a risky diff worth an outside review, or a tie between two interpretations you cannot break from the evidence alone. The user may run it explicitly (/consult-codex), but also self-invoke it — without being asked — when an outside perspective would actually change the risk on a hard call. Skip it for trivial lookups, routine edits, and decisions you are already confident in: each consult spends real tokens and a few seconds, so spend it only where cross-model divergence would genuinely change the outcome.From its SKILL.md

Install
npx -y skills add adeilsonrbrito/cross-model-consult --skill consult-codex

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.3 KB, ~1.1k tokens by cl100k_base, as published. Nobody here has run it

Consult Codex

Why this exists

Codex runs a different model (and provider) than you do. When you ask it the same question, agreement is a strong signal you're on solid ground, and disagreement surfaces a blind spot you couldn't see from inside your own reasoning. This is the same principle behind adversarial multi-model debate: divergence between genuinely different reasoning systems is information. A single model agreeing with itself is not.

Use it to triangulate a hard decision, pressure-test a design before building, get a review of a diff or a plan, or break a tie when you're stuck between two interpretations. It is not for trivial lookups or things you're already sure of — each call spends real tokens and a few seconds, so spend it where an independent perspective actually changes the risk.

How to consult Codex

1. Write a self-contained prompt — Codex sees none of your conversation

This is the most important step. Codex starts cold: it does not see this chat, the files you've read, or the decision history. Whatever it needs to reason well, you must put in the prompt. Write it to a file (avoids shell-quoting pain) and include:

  • Context — the relevant facts, code, constraints, and the options on the table. Enough that Codex can reason, not so much that you bury the question.
  • The specific question — one clear ask, with the concrete alternatives if it's a fork.
  • A standing instruction to disagree if warranted — e.g. "You are an independent reviewer. Reason from the facts below and give a decisive recommendation. Do not just agree — if you disagree, say so and why." You want its real opinion, so present the context fairly and don't lead it to your conclusion.

Write the prompt to a temp file, e.g. /tmp/codex-query.md.

2. Run it through the wrapper script

bash <skill-dir>/scripts/ask-codex.sh /tmp/codex-query.md

The script encapsulates the exact non-interactive invocation (trust-gate override, approval policy, read-only sandbox, stdin piping) so it works first time. Read the header comments in scripts/ask-codex.sh for what each flag does and why it can't be dropped. Optional overrides via env var:

CODEX_MODEL=gpt-5.5 bash <skill-dir>/scripts/ask-codex.sh /tmp/codex-query.md
# CODEX_SANDBOX=workspace-write   # only if Codex must write/run; default read-only is right for advice

If you ever need the raw command (no script), it is:

codex -a never -c 'projects."<ABSOLUTE_CWD>".trust_level="trusted"' exec -s read-only - < /tmp/codex-query.md

3. Report back, and surface divergence honestly

Relay Codex's verdict, then say plainly where it agrees and where it diverges from your own view — don't quietly fold its answer into yours, and don't quietly ignore it. If Codex contradicts a conclusion you'd already reached from evidence, don't just switch sides: name the conflict and reconcile it ("I found X, Codex argues Y, here's which constraint breaks the tie"). Cross-model agreement is the product; so is a well-explained disagreement. Clean up the temp prompt file when done.

Interpreting the result

  • Both agree → strong signal. State it; it's earned confidence, not just repetition.
  • Codex disagrees → the valuable case. Weigh its reasoning on the merits. If it has a point you missed, adapt. If you have primary evidence it underweighted, hold your ground and say why. Either way the user sees both sides.
  • Treat Codex as a peer reviewer, not an oracle. It can be confidently wrong; so can you.

Troubleshooting

  • Not inside a trusted directory ... --skip-git-repo-check was not specified — the trust-gate override didn't match the cwd. The script uses $(pwd); make sure you're invoking from the directory whose context you want Codex to assume.
  • The call hangs — almost always a missing -a never (an approval prompt is waiting on a TTY that isn't there). The script includes it; don't strip it.
  • codex: command not found / wrong binary — verify with codex --version. There's a name collision with an unrelated tool (reachingforthejack/rtk ships its own rtk); the same caution applies if multiple codex-like tools are installed.
  • Need Codex to read repo files to answer — read-only already permits reading files inside the cwd. Only escalate CODEX_SANDBOX if it must write or run commands.

What ships with it: 1 file

1.8 KB alongside SKILL.md, 1 of them executable

scripts/

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.