context-optimization

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOW
Full Analysis
  • SAFE (SAFE): No security findings identified. The file consists of documentation, architectural concepts, and non-functional code examples intended to guide agent behavior for token efficiency.
  • Indirect Prompt Injection (INFO): While the skill describes processing untrusted external data (tool outputs and retrieved documents), it does not provide active tools or scripts to perform these operations, nor does it define capabilities that could be exploited via injection. The risk is limited to internal agent reasoning logic.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 10:56 AM