claude-code-review
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill utilizes standard, well-known developer tools including
git,gh(GitHub CLI), and theclaudeCLI for code analysis and project interaction. - [SAFE]: The instructions incorporate specific security controls to prevent command injection, such as advising the use of absolute paths, shell quoting for user-provided values, and single-quoted heredocs (
<<'CLAUDE_PROMPT') to prevent variable expansion during shell execution. - [SAFE]: A strict read-only constraint is enforced, explicitly instructing the agent not to modify or create files, which limits the potential impact of any processing errors or adversarial input during the review process.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface as it ingests untrusted data from pull requests, branches, and source files. However, it mitigates this risk by clearly separating the untrusted code content from the logic of the review process and employing restricted execution patterns.
Audit Metadata