What problem does it solve?
Broad life-sciences questions often span multiple databases and evidence types, making it hard to know which sources to query and how to combine results. This Skill acts as an orchestration layer that interprets open-ended research questions, normalizes key entities, selects the smallest useful set of downstream skills, and synthesizes a concise evidence-backed answer.
Core Features & Use Cases
- Research Task Classification: Sorts requests into lanes such as human genetics, variant interpretation, pathway biology, chemistry and pharmacology, clinical evidence, and literature discovery.
- Entity Normalization: Resolves genes, diseases, variants, compounds, pathways, and dataset identifiers to the right lookup skills before deep retrieval.
- Skill Selection Heuristics: Maps common question types (target-disease evidence, variant interpretation, locus-to-gene mapping, structure and mechanism) to minimal skill sets like Open Targets, ClinVar, gnomAD, ChEMBL, and Reactome.
- Subagent Parallelization: Delegates independent evidence lanes to subagents when available, while keeping interpretation, conflict resolution, and final synthesis centralized.
- Use Case: A user asks "what is known about the association between gene X and disease Y?" The router normalizes the gene and disease, gathers genetics, expression, and clinical evidence in parallel, and returns a structured answer with caveats and next steps.
Quick Start
Ask a broad life-sciences research question such as what is known about a gene's role in a disease and have the router plan the retrieval and synthesize the evidence.
