requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Jul 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of shell commands like
git rev-parseandgit logto identify commit ranges. Additionally, the subagent template incode-reviewer.mdinstructs the subagent to usegit diff,git show, andgit worktreeto inspect the codebase history and state. - [PROMPT_INJECTION]: The skill processes untrusted code changes which creates a potential surface for indirect prompt injection attacks. 1. Ingestion points: The subagent ingests output from
git diffcontaining raw code and developer comments incode-reviewer.md. 2. Boundary markers: The prompt template lacks explicit instructions or delimiters to warn the subagent against following instructions embedded within the reviewed code. 3. Capability inventory: The subagent is granted shell access for git operations and filesystem access. 4. Sanitization: The skill does not sanitize or filter the content of the git diffs before processing.
Audit Metadata