research-router-skill

Pass

Audited by Gen Agent Trust Hub on Apr 20, 2026

Risk Level: SAFE
Full Analysis
  • Orchestration Design: The skill provides a logical framework for routing research questions to appropriate specialized skills (e.g., genetics, pharmacology, clinical evidence). This structural approach is a standard functional pattern for multi-agent or multi-tool workflows and does not perform any dangerous actions.
  • Entity Normalization: The instructions guide the agent to resolve and normalize scientific entities (genes, diseases, compounds) using specific sub-skills before proceeding with retrieval. This ensures data consistency and reduces the risk of processing ambiguous or malformed input.
  • Least Privilege and Efficiency: The skill includes heuristics to select the "smallest useful set of skills" and prefers direct lookups over expensive chains, which aligns with best practices for resource management and minimized capability exposure.
  • Data Handling: The skill processes life-sciences data and synthesizes evidence-backed answers. It does not contain instructions for accessing sensitive local files, environment variables, or hardcoded credentials.
  • Subagent Coordination: The guidelines for using subagents emphasize read-only objectives and bounded scopes, which provides a safe mechanism for parallelizing independent data gathering tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 20, 2026, 02:27 PM