skills/quad4-software/ai/memory/Gen Agent Trust Hub

memory

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is storing and retrieving arbitrary text data through remember and recall tools. This introduces a vulnerability surface where a user or an external process could feed malicious instructions into the agent's memory. When the agent later retrieves these 'memories,' the LLM may treat the stored text as a new set of instructions, potentially bypassing original system constraints.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 10:56 PM
Security Audit — agent-trust-hub — memory