memory-systems
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill functions as an educational guide for designing memory architectures. It does not include executable scripts, shell commands, or instructions that would compromise a system.
- [EXTERNAL_DOWNLOADS]: The skill references well-known technology services such as Neo4j, Pinecone, and Weaviate. These are documented neutrally as resources for implementing the described architectures and do not involve untrusted or risky downloads.
- [INDIRECT_PROMPT_INJECTION]: The skill addresses the implementation of systems that process historical data, which inherently involves an indirect prompt injection surface. However, it explicitly includes security guidelines recommending that developers treat retrieved memories as "untrusted historical evidence" and implement tenant/access filters, which follows security best practices.
- [COMMAND_EXECUTION]: The provided Python and Cypher examples are benign pseudocode intended for demonstration purposes and do not invoke dangerous system-level commands or unauthorized file access.
Audit Metadata