codex-crosscheck

A command that asks OpenAI Codex for a second opinion and has another agent compare that answer with the code. It returns the question, both positions, verification notes, and a synthesis.

In plain words
What is it for?
Use it to review design choices, debugging questions, or other coding decisions when an independent answer and code-based cross-check are useful.
Why use it?
It helps expose disagreements or unsupported claims before making a technical decision. The code comparison distinguishes what was measured, inferred, or left open.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/fidpa/codex-sandbox-toolkit/codex-crosscheck
Clone the repo
git clone --depth 1 https://github.com/fidpa/codex-sandbox-toolkit
Per session 25 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 404 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00025 $0.00404
Opus 5 $0.00013 $0.00202
Sonnet 5 $0.00005 $0.00081
Haiku 4.5 $0.00003 $0.00040

Measured 2d ago against content hash fcd53b5a3ddb, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

codex-crosscheck scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

claude/commands/codex-crosscheck.md · 49 lines

What it actually says

Codex cross-check

Description: Delegate the question to the codex-advisor subagent. It fetches the Codex answer, verifies its claims against the code, and delivers a synthesis against its own position.

Argument: $ARGUMENTS: the question. If it is missing, ask once, briefly.

Costs more than /codex: on top of the Codex quota it also spends Claude tokens for the shell around it. If you only want the OpenAI answer, use /codex.


🚀 EXECUTION DIRECTIVE

Invoking this slash command IS the explicit user request. Start IMMEDIATELY, NO follow-up questions (unless the question itself is missing).


Procedure

  1. Task the codex-advisor subagent with the question from $ARGUMENTS. Do not call the wrapper yourself; that would be /codex.

  2. Give the subagent the context it needs: which files are involved and what the decision is. It forms its own position from that before asking Codex.

  3. Relay its synthesis. It contains four parts: the question asked, its own position, Codex's position with a note on what was verified, and the synthesis separating measured, inferred and open.

  4. Do not smooth over disagreement between the models. The dissent is the result, not a flaw in it.

When this command is the right one

  • Architecture decisions that would be expensive to reverse
  • Root-cause analysis after two fixes have failed
  • Security questions: authentication, cryptography, untrusted input
  • Non-trivial algorithms with edge cases

Not for naming questions, style, or understanding code. And not when you simply want the OpenAI answer.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 49 lines · 25 tokens per session scan A fcd53b5a3ddb

Subscribe to this mod's changes

codex-crosscheck is a command published in the GitHub repository fidpa/codex-sandbox-toolkit (1 stars, last pushed 3d ago), licensed MIT. It adds 25 tokens to every session and 404 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.