codex-code-review
Pass
Audited by Gen Agent Trust Hub on Feb 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the execution of multiple command-line tools to function. It invokes
codex --full-auto cto generate code reviews, performs Git operations likegit addandgit committo manage changes, and uses abacklogCLI tool to manage tasks based on findings. It also executes a local helper scriptscripts/parse_codex_review.shto process review data. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it parses and acts upon suggestions found in externally generated review files.
- Ingestion points: Reads markdown files from the
.agent/reviews/directory generated by thecodexCLI. - Boundary markers: No explicit delimiters or warnings are used to separate the review content from the agent's logic.
- Capability inventory: The agent is authorized to modify source code, commit changes via Git, and interact with a backlog system.
- Sanitization: No validation or sanitization is performed on the content of the 'Suggested fix' fields before the agent implements them in the codebase.
Audit Metadata