context-engineering
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- SAFE (SAFE): The skill provides purely instructional content and architectural guidelines for managing LLM context windows. All analyzed content is documentation-based.
- Code Snippets as Documentation (SAFE): Python code blocks found in the reference files (e.g., in
references/context-degradation.mdandreferences/memory-systems.md) are illustrative examples of logic rather than executable scripts intended for direct deployment. They do not contain any malicious behavior. - No Sensitive Data (SAFE): No hardcoded credentials, API keys, or sensitive file paths were identified. The architectural patterns suggested follow standard software development practices (e.g., using a local
/datadirectory for state management). - Reference-Only External Tools (SAFE): The skill mentions established external tools and libraries like Zep, MemGPT, and GraphRAG in a comparative/educational context without providing commands to download or execute them from untrusted sources.
Audit Metadata