context-degradation
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill provides documentation on LLM behavior patterns like 'Lost-in-Middle' and 'Context Poisoning'. It contains no code, shell commands, network operations, or sensitive data access. The content is designed to assist with diagnosing agent performance issues and does not include any instructions intended to bypass safety guardrails or persist across sessions.
Audit Metadata