vector-specialist

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Prompt Injection] (SAFE): No instructions to bypass safety filters or extract system prompts were found. The skill instructions are focused exclusively on domain expertise.
  • [Data Exposure & Exfiltration] (SAFE): No hardcoded credentials or unauthorized network calls detected. The snippets use standard environmental variables for configuration.
  • [Remote Code Execution] (SAFE): Code examples use legitimate high-level library APIs (Qdrant, Redis) and do not involve dynamic execution of untrusted code.
  • [Indirect Prompt Injection] (SAFE): Although the query expansion pattern ingests user input into an LLM call, it is restricted to generating search variations and is a standard retrieval optimization technique.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:14 PM