engram-memory-protocol
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions define a protocol for utilizing internal memory tools to persist project context. It emphasizes storing verified, durable knowledge like rationales and root causes.
- [SAFE]: It includes a specific security-positive instruction to exclude secrets, private data, and speculation from persistent storage.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external inputs which presents an attack surface. 1. Ingestion points: User-provided decisions, bugfixes, and discoveries (SKILL.md). 2. Boundary markers: None specified. 3. Capability inventory: Uses memory tools (mem_save, mem_update) for persistence. 4. Sanitization: Explicit instructions to exclude secrets and private data. The presence of sanitization instructions mitigates the risk level.
Audit Metadata