git:20260724.a7a537f to git:20260729.c6636b9

11 added, 12 removed. Audit A to A.

---

name: codex-advisor

description: 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".

---

# Codex Advisor

Get a focused second opinion from GPT through the Codex CLI, without handing the

work over to it.

## Route by tool

1. In Claude Code, delegate to the native `codex-advisor` agent. It receives the

recent conversation automatically, so do not paste the history yourself.

- 2. In Codex, Cursor, or Gemini CLI, locate this skill's directory and run

- `node scripts/ask_codex.mjs`. Pass the review request on standard input.

+ 2. Elsewhere, run `node scripts/ask_codex.mjs` from this skill's directory with

+ the review request on standard input.

- The request must include the decision or conclusion, the relevant recent

- context and evidence, and any constraints or alternatives that affect the

- verdict. Include only facts that matter to the decision.

+ State the decision, the evidence behind it, and any constraint that changes the

+ verdict. Name the paths that matter: the reviewer has read-only access and

+ checks load-bearing claims itself.

- Return the reviewer's answer without rewriting it. If `codex` is missing or not

- authenticated, surface the command error and ask the user to install the Codex

- CLI or sign in. Do not fall back to the host model.

+ Return the answer without rewriting it. If `codex` is missing or unauthenticated,

+ surface the error and ask the user to install it or sign in. Never review in its

+ place.

- ## Picking the reviewer

+ ## Reviewer

- The review runs on whatever model `~/.codex/config.toml` sets. In Codex itself

- that is the same model already doing the work, which is a weak second opinion,

- so set `CODEX_ADVISOR_MODEL` to a different one before calling.

+ Pinned to `gpt-5.6-sol` at medium reasoning effort. Inside Codex that may be the

+ model already doing the work, so set `CODEX_ADVISOR_MODEL` to a different one.

## Weighing the answer

Give the verdict serious weight. If a step it recommends fails when tried, or a

file contradicts a specific claim, surface the conflict instead of following the

review blindly.