agent-memory

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and retrieval of persistent data into the agent's context window, which creates a surface for indirect prompt injection if the stored content is influenced by untrusted inputs.
  • Ingestion points: The skill reads data from local markdown files in the memory/ directory via FileMemory.search and from database storage in the MemoryStore.search (SQLite) and ChromaMemory.recall (ChromaDB) methods.
  • Boundary markers: The provided code does not implement explicit delimiters or boundary markers for the retrieved context, though guidelines recommend prepending memories to the system prompt rather than user messages.
  • Capability inventory: The skill has the capability to read and write files to the local file system and interact with SQLite and ChromaDB databases.
  • Sanitization: No sanitization or content validation is performed on retrieved memories before they are injected into the agent's prompt context.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 03:43 PM
Security Audit — agent-trust-hub — agent-memory