The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill has an attack surface for indirect prompt injection because it is designed to analyze and modify external code data. • Ingestion points: 'recently modified code' referred to in the skill logic. • Boundary markers: Absent; there are no instructions to delimit the code from the agent's core instructions. • Capability inventory: The skill is capable of modifying and replacing existing code. • Sanitization: None provided for the code content being processed.
[Safe Behavior] (SAFE): No indicators of data exfiltration, hardcoded credentials, persistence mechanisms, or unauthorized network activity were detected. The reference to 'http://CLAUDE.md' appears to be a documentation convention rather than a malicious URL.

simplify