storm-research

Invoke this Skill with a research question and intended deliverable. It frames the topic, builds a sourced baseline, dispatches one isolated executor per lens, then synthesizes the returns into a structured briefing with citations.

STORM is Stanford OVAL's method of synthesizing topic outlines through retrieval and multi-perspective question asking. This Skill adapts that pattern as an independent research workflow using canonical lenses like Practitioner, Skeptic, and Economist; it is not a distribution of the STORM software.

Yes. The user can narrow the canonical lens set, and the Skill records the narrowing before dispatch. Topic-specific lenses such as regulator, customer, or adversary can be added when omitting them would change the answer.

It maps contradictions explicitly, keeping each conflicting conclusion with its evidence rather than replacing disagreement with a vague hedge. Key tensions appear in the briefing with evidence on each side and what remains unresolved.

The run is marked degraded: factual work is labeled an unverified perspective simulation or single-context synthesis, confidence is lowered, and the limitation plus what would upgrade the result is recorded and disclosed.

Avoid it for quick factual lookups or short answers; the description restricts it to deliverables that are full research briefings, STORM-style investigations, or evidence reviews. The multi-lens dispatch and fidelity review add overhead unjustified by simple questions.