code-review
Pass
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides legitimate instructions for performing code reviews and refactoring. No obfuscation, hardcoded credentials, or malicious persistence mechanisms were detected.
- [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection as it processes external documentation to guide its actions. 1. Ingestion points: Reads implementation criteria from
.agents/issues/spec.md. 2. Boundary markers: Absent; the skill lacks specific delimiters or instructions to ignore embedded prompts within the specification file. 3. Capability inventory: The skill is authorized to perform file writes ("Fix identified smells") and command execution ("verify test suite"). 4. Sanitization: No sanitization or validation of the specification content is defined.
Audit Metadata