caching
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions involve analyzing application logs (log/development.log) and querying the cache database (e.g., SolidCache::Entry) to identify bottlenecks or inspect cache state. Since logs and databases can contain untrusted user-controllable data, they serve as potential ingestion points for indirect prompt injection. This is a standard surface for performance optimization tools.
- Ingestion points: log/development.log and the application database.
- Boundary markers: None explicitly enforced in the instructions.
- Capability inventory: Access to Read, Write, Edit, and Bash execution.
- Sanitization: No explicit sanitization of external data is mentioned.
Audit Metadata