oh-notes

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests and processes untrusted pull request comments.
  • Ingestion points: Pull request comments and reviews are retrieved from GitHub using gh pr view and the GitHub API in SKILL.md.
  • Boundary markers: There are no explicit delimiters or instructions to the agent to treat the fetched comments as external, untrusted content.
  • Capability inventory: The skill has the authority to perform file system modifications, git push operations, and create GitHub issues and replies.
  • Sanitization: While sg review is executed on staged changes, this tool is described as a code review utility rather than a security filter designed to detect or sanitize malicious instructions embedded in natural language comments.
  • [SAFE]: The skill communicates with GitHub, a well-known service, for repository management and feedback retrieval.
  • [SAFE]: Command execution is limited to standard version control and CLI tools (git, gh) used for their intended purpose in a development workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 05:14 PM