code-reviewer

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • COMMAND_EXECUTION (MEDIUM): The skill automatically executes the yq command during SessionStart to read .claude/config.yaml.
  • Evidence: command: if command -v yq && [ -f ".claude/config.yaml" ]; then yq -o=json '.constraints' .claude/config.yaml.
  • Risk: Processing untrusted project files with system tools can lead to behavior manipulation if the configuration is malicious.
  • PROMPT_INJECTION (HIGH): The skill is vulnerable to Indirect Prompt Injection (Category 8) because it ingests and processes untrusted external data (code) while having command execution capabilities.
  • Ingestion points: Untrusted source code files provided for review and .claude/config.yaml.
  • Boundary markers: No explicit delimiters or 'ignore instructions' warnings are present in the methodology.
  • Capability inventory: Shell command execution via yq.
  • Sanitization: No sanitization or validation of external code content or configuration data is implemented.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 04:56 AM