discrete-mathematics-expert

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill primarily consists of mathematical theory, pedagogical guidance, and LaTeX-formatted formulas. No suspicious behavior, network exfiltration, or unauthorized file access was identified.
  • [NO_CODE]: The skill contains no executable scripts (Python, JavaScript, shell scripts) or complex logic, significantly limiting the technical attack surface.
  • [PROMPT_INJECTION]: The research workflow in modules/research-checklist.md instructs the agent to ingest external data from Wikipedia and Arxiv using specific tools. While this constitutes an attack surface for indirect prompt injection, it is considered low risk given the focus on academic sources.
  • Ingestion points: External data retrieved from Wikipedia and Arxiv via adn_skills tool calls.
  • Boundary markers: Absent; the skill does not specify delimiters for wrapping the ingested external content.
  • Capability inventory: No file-write, subprocess execution, or direct network operations are defined within the skill's own content.
  • Sanitization: None; the skill relies on the agent's default processing of fetched research data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 05:38 PM
Security Audit — agent-trust-hub — discrete-mathematics-expert