Skills
Navigate
skills/rlaope/oh-my-hermes/omh-code-review/Gen Agent Trust Hub

omh-code-review

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as pull requests, code diffs, and project artifacts.\n
  • Ingestion points: Input includes diffs, source files, and artifacts typically located in .omh/artifacts/ or .omh/handoffs/.\n
  • Boundary markers: The skill employs an "evidence boundary" and requires "checked-and-clean" and "could-not-assess" lists to delimit assessed surfaces.\n
  • Capability inventory: The skill interacts with the critic harness for runtime recording and utilizes shell tools for repository state inspection.\n
  • Sanitization: The references/review-dispatch.md file mandates verifying reports against command output and explicitly warns "Do not trust the report," establishing a skeptical stance toward input claims.\n- [COMMAND_EXECUTION]: The skill uses shell commands to identify the specific git range for review.\n
  • Evidence: Usage of git merge-base and git rev-parse in references/review-dispatch.md as well as omh runtime record in SKILL.md for session tracking.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 08:33 PM
Security Audit — agent-trust-hub — omh-code-review