skills/mitodl/agent-kit/code-review/Gen Agent Trust Hub

code-review

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code content from git diffs and GitHub Pull Requests. This input is interpolated into the agent context without explicit boundary markers, allowing instructions within the reviewed code to potentially manipulate agent behavior.
  • Ingestion points: External code data enters the context through git diff and gh pr diff commands in SKILL.md.
  • Boundary markers: The instructions do not define delimiters or instructions to ignore commands within the reviewed code.
  • Capability inventory: The agent can read files, execute git/gh commands, and modify the filesystem in 'fix mode'.
  • Sanitization: No input validation or filtering of the code content is specified.
  • [COMMAND_EXECUTION]: The skill generates shell commands using git and gh based on user-provided branch names and PR numbers. While functional, this presents a command injection surface if inputs are not sanitized.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 12:31 PM
Security Audit — agent-trust-hub — code-review