agent-memory
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill is comprised of markdown documentation and design patterns for building memory systems. No executable code or malicious patterns were found.
- [INDIRECT_PROMPT_INJECTION]: The documentation identifies and addresses the risk of indirect prompt injection where memory could carry malicious instructions from external sources. 1. Ingestion points: Background extraction from finished conversation messages (WRITE.md). 2. Boundary markers: Recommends limiting agent capabilities and treating user preferences as read-only (SKILL.md). 3. Capability inventory: Memory retrieval can influence agent tool-calling and response generation (READ.md). 4. Sanitization: Advocates for scanning and redacting secrets, card numbers, and PII before storage (LIFECYCLE.md).
- [NO_CODE]: The provided files are purely informational and do not include any scripts, executable tools, or automated execution logic.
Audit Metadata