Review code diffs via OpenAI Codex CLI with pass/fail verdicts.
npx skills add https://github.com/Intension-us/attention-diagnostic --skill codex-intension-us
Please help me install this Agent Skill. Skill: codex Source: https://github.com/Intension-us/attention-diagnostic/tree/main/.claude/skills/gstack/codex Command: npx skills add https://github.com/Intension-us/attention-diagnostic --skill codex-intension-us
OpenAI Codex CLI wrapper — three modes. Code review: independent diff review via codex review with pass/fail gate. Challenge: adversarial mode that tries to break your code. Consult: ask codex anything with session continuity for follow-ups.
Invoke /codex review to get an independent second opinion on your latest diff.
High-intent search queries and answers about installing and using this skill.
To get an independent second opinion on a code diff, invoke the codex review mode to analyze changes and return a pass/fail verdict. It operates as a strict, adversarial code review gate using the OpenAI Codex CLI.
Adversarial code review is a challenge mode that actively tries to break your code logic and find edge cases. It operates by analyzing your diff or prompt against project context to return a precise verdict and cost estimate.
Yes, you can ask follow-up questions using consult mode. It provides session continuity, allowing you to ask the codex anything with context preserved from previous interactions while enforcing strict safety boundaries.
Yes, the codex CLI wrapper enforces filesystem boundaries during code analysis. It reads project context safely by applying a strict prompt boundary and telemetry hooks to preserve system security.
Limitations of using an automated code review gate include relying entirely on the OpenAI Codex CLI's understanding of your project context. While it provides brutally honest verdicts, complex architectural flaws may exceed its diff analysis scope.
ruah-devLoad repository and environment context from filesystem configuration and recent commits.
him55710-sudoReview code diffs via Codex CLI and return a gate verdict.
hongbietcodeCreate and configure deterministic Claude Code hooks for tool lifecycle events.
Chang-Shih-YungAggregate type checker, linter, and test results into a 0-10 composite score.
