engram-memory-protocol
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructions create a surface for indirect prompt injection by directing the agent to use user-provided keywords for memory retrieval.
- Ingestion points: User references to projects, features, or problems in the initial session message (SKILL.md).
- Boundary markers: No specific delimiters or instructions are provided to isolate user input from the memory tool parameters.
- Capability inventory: The skill utilizes memory tools including mem_save, mem_search, mem_context, and mem_session_summary.
- Sanitization: There is no instruction to validate or sanitize user keywords before they are used in search queries.
Audit Metadata