agent-memory
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to store and recall "user-stated rules" and "lessons" which are intended to influence future agent behavior. While this is the primary purpose of the skill, it creates a surface for indirect prompt injection if malicious instructions are recorded as rules.
- Ingestion points: The agent reads instructions and state from
wiki/memory/MEMORY.mdat the start of tasks. - Boundary markers: No specific delimiters or safety instructions are defined to distinguish memory content from system instructions during recall.
- Capability inventory: The skill involves reading and writing workspace files and suggests the potential for shell command execution (subject to user permission).
- Sanitization: The instructions do not specify validation or sanitization for the content stored in the memory file.
Audit Metadata