vector-memory

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill stores memory records locally at ~/.thepopebot/vector-memory.json. This file system access is consistent with the described functionality of providing persistent semantic memory.
  • [EXTERNAL_DOWNLOADS]: The skill uses the official openai Node.js library to communicate with OpenAI's API for generating text embeddings. This is a standard and necessary integration for the skill's primary purpose.
  • [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection.
  • Ingestion points: Untrusted data enters the agent context through the vstore function in vector-memory.js.
  • Boundary markers: None; text is stored and retrieved without specific delimiters or warnings to the agent.
  • Capability inventory: The skill does not possess dangerous capabilities like arbitrary command execution or network requests beyond the OpenAI API.
  • Sanitization: No sanitization is performed on the stored text, which is typical for memory-storage utilities.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 05:11 AM