memory-notes
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists entirely of instructional documentation for a memory management system. It contains no executable scripts, remote code execution patterns, or unauthorized network operations. All resource names and tools mentioned are consistent with the Basic Memory ecosystem.
- [INDIRECT_PROMPT_INJECTION]: The skill describes a framework for processing and maintaining a knowledge graph from memory notes. This creates an ingestion surface where an agent processes external content. While the skill is safe, this represents a pathway for potential indirect prompt injection if an agent reads notes from untrusted sources.
- Ingestion points: Notes are retrieved into context via the search_notes, read_note, and build_context tools.
- Boundary markers: Note anatomy includes structured sections like frontmatter and headers which provide delimiters but are not security-focused boundaries.
- Capability inventory: Tools like write_note, edit_note, and move_note allow the agent to modify note content.
- Sanitization: No sanitization methods are described in the instructions to filter note content for embedded malicious instructions.
Audit Metadata