memory-systems
Pass
Audited by Gen Agent Trust Hub on Apr 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a technical resource for implementing agent persistence. No malicious patterns, such as prompt injection, data exfiltration, or obfuscation, were detected across the documentation or implementation scripts.
- [EXTERNAL_DOWNLOADS]: The skill references and provide examples for integrating well-known memory frameworks and libraries, including numpy, mem0, graphiti_core, and cognee. These references are provided for educational and comparative purposes and align with the skill's stated purpose.
- [PROMPT_INJECTION]: The skill provides reference code for retrieving and interpolating memories into an agent's context (e.g., the MemoryContextIntegrator class). While systems that ingest external data are theoretically subject to indirect prompt injection, the implementation follows standard architectural patterns and includes guidelines for managing memory quality and temporal validity to mitigate common issues.
Audit Metadata