omh-code-review
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted content such as code diffs and pull request descriptions. To mitigate the risk of indirect prompt injection, the
references/review-dispatch.mdfile includes explicit instructions for the agent to report any commands or instructions found within the reviewed material as findings rather than executing them. - [COMMAND_EXECUTION]: The skill documentation describes the use of
git,gh(GitHub CLI), andomh(platform-specific tool) for obtaining code changes and recording runtime evidence. These tools are used within their intended scope for software development and auditing workflows. - [DATA_EXPOSURE]: Analysis of
SKILL.mdandreferences/architecture-delta.mdshows that the skill reads file contents, diffs, and session metadata. This data access is necessary for the stated purpose of code review and is confined to the local project environment with no evidence of unauthorized exfiltration.
Audit Metadata