code-review

Warn

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill suggests cloning a repository from an unverified GitHub user ('truffi/shellcheck') for use as a security scanner, which poses a risk of executing untrusted code.
  • [COMMAND_EXECUTION]: Provides instructions to run local tools and scripts such as 'npm test', 'npm run lint', and 'sonar-scanner'. These commands execute logic defined in the project's 'package.json' or local configuration, which can be exploited to run arbitrary code if the repository being reviewed is malicious.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests untrusted data from 'git diff' and file contents. Ingestion points: Git diff outputs and file search results. Boundary markers: None present. Capability inventory: Shell command execution (npm/git) and GitHub API access. Sanitization: No escaping or validation of external content is specified.
  • [CREDENTIALS_UNSAFE]: Includes commands to search for and expose hardcoded secrets, passwords, and API keys ('grep -rE "password|secret|api[_-]?key"'). While intended for auditing, this behavior extracts sensitive information into the agent's active context.
  • [REMOTE_CODE_EXECUTION]: The combination of cloning an unverified repository and the potential to execute its contents constitutes a remote code execution risk.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 1, 2026, 05:10 AM