codex-review
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires installation from a third-party GitHub repository (
BenedictKing/codex-review) which serves as the source for thenpx skills addcommand. - [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to review code, which involves reading untrusted files from a user's workspace. This creates an attack surface where instructions embedded in code could influence the agent.
- Ingestion points: User code files read during the review process (referenced in SKILL.md).
- Boundary markers: Absent; the instructions do not specify using delimiters or specific prompting techniques to separate code content from instructions.
- Capability inventory: The skill description implies the agent will have file system access for reading code and writing/modifying
CHANGELOG.mdfiles. - Sanitization: No mention of input validation or sanitization for potentially malicious content in code comments or metadata.
Audit Metadata