memory-hygiene
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill manages persistent memory stores, creating a surface for indirect prompt injection if malicious instructions are stored and later retrieved.
- Ingestion points: Memory retrieval and search operations described in the 'Retrieval' section (SKILL.md).
- Boundary markers: The skill advises using 'untrusted-input-hygiene' for embedded attempts to redirect tasks.
- Capability inventory: Reading and writing to file-backed notes and managed memory services.
- Sanitization: Instructions recommend treating notes as evidence rather than instructions and applying hygiene procedures to retrieved content.
- [SAFE]: The skill explicitly advises against storing credentials, tokens, or secrets in memory, which is a positive security practice.
Audit Metadata