context-compression
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOW
Full Analysis
- [SAFE] (SAFE): The file 'references/evaluation-framework.md' is a static documentation file. It specifies scoring rubrics, probe types, and LLM judge configurations for evaluating context compression. There are no triggers for command execution, network activity, or data exfiltration. The included system prompt for an LLM judge is a standard instructional template for evaluation and does not constitute a prompt injection attack against the agent itself. No dependencies or remote code patterns were identified.
Audit Metadata