Wraps the OpenAI Codex CLI to run independent code reviews, adversarial challenges, and consult sessions.
npx skills add https://github.com/sanjanb/my-agent-harness --skill codex-sanjanb
Please help me install this Agent Skill. Skill: codex Source: https://github.com/sanjanb/my-agent-harness/tree/main/skills/codex Command: npx skills add https://github.com/sanjanb/my-agent-harness --skill codex-sanjanb
codex review on your diff with a pass/fail gate, giving an outside perspective before you ship. - Adversarial Challenge Mode: Tries to break your code by attacking assumptions, edge cases, and design decisions. - Consult with Session Continuity: Ask Codex anything and keep follow-up context across questions. - Use Case: Before merging a risky refactor, invoke the skill to have Codex review the diff, then run challenge mode to probe for failure modes you missed. ## Quick Start Ask the agent to run a codex review on the current branch diff and report any blocking issues.High-intent search queries and answers about installing and using this skill.
Invoke the codex skill's review mode, which runs `codex review` on your current diff and returns findings with a pass/fail gate. It acts as an independent reviewer outside your main agent's reasoning.
Adversarial review tries to break your code by attacking assumptions and edge cases rather than just checking style. Use the skill's challenge mode, triggered by phrases like "codex challenge" or "outside voice challenge".
Yes, consult mode maintains session continuity so you can ask follow-up questions without re-establishing context. This is useful for multi-turn design discussions with the Codex CLI.
The skill detects headless sessions via its preamble and adjusts behavior, blocking on interactive prompts when no human can answer. Interactive prompts are skipped or auto-decided in spawned sessions.
The review quality depends on the Codex CLI being installed and authenticated, and results reflect a single model's perspective. It complements but does not replace human review for high-risk changes.
fwojciecAutomate code reviews by querying Gemini on Git branch diffs.
ShaheerKhawajaChain self-evaluation, review, and ship steps to validate code before deployment.
dorrioFlag tests validating external framework behavior instead of custom application code.
saffron-healthRender session diffs as a Markdown walkthrough in a native Glimpse window.