code-simplifier
Fail
Audited by Gen Agent Trust Hub on Feb 15, 2026
Risk Level: HIGHPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (HIGH): The skill is designed to ingest and refine external code, creating an attack surface where malicious instructions could be embedded in code or comments.\n
- Ingestion points: The skill instructions specify analyzing 'recently modified code' and 'code sections' as input (SKILL.md).\n
- Boundary markers: The instructions provide no delimiters or 'ignore' warnings to separate the logic of the code being analyzed from the agent's own instructions.\n
- Capability inventory: The skill is tasked with 'applying refinements' and 'simplifying code structure' autonomously, which constitutes a write capability on the codebase.\n
- Sanitization: No sanitization, escaping, or validation logic is defined to prevent the agent from being influenced by natural language instructions found within the code blocks it processes.
Recommendations
- AI detected serious security threats
Audit Metadata