Codex Code Review now supports custom repository rules defined in AGENTS.md, allowing teams to codify reviewer preferences and catch repository-specific issues even when context is siloed.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
OpenAI announces that Codex Code Review now supports defining custom repository rules via AGENTS.md files, enabling AI reviews to catch repository-specific issues even when relevant context is isolated.
On July 21, 2026, OpenAI Developers announced on X that Codex Code Review now supports defining custom repository rules via AGENTS.md files. This update allows teams to convert checkpoints that reviewers repeatedly emphasize into rules, enabling Codex to identify repository-specific issues even when relevant context is isolated.
OpenAI recommends that users start with rules that reviewers repeatedly check, and keep rules concise and clearly scoped. This way, Codex can effectively catch repository-specific issues even when context is isolated.
Information comes from OpenAI's official X account, a first-party release, so credibility is relatively high. However, specific implementation details and effectiveness have not yet been verified by third parties.
Codex Code Review supports custom rules via AGENTS.md, making AI reviews more aligned with team standards and improving the relevance and efficiency of code reviews.
Primary report
Primary source