storm-research

name

storm-research

description

Use for non-trivial research, strategic decisions, market/company/topic analysis, briefings, due diligence, or any request that needs more than a single-source answer. Runs a STORM-style multi-perspective workflow with source gathering, expert lenses, contradiction mapping, synthesis, confidence review, and citations. Trigger when the user asks for a research brief, deep research, "look at this from multiple angles", compare evidence, find blind spots, prepare for a decision, or use the Research Bee's STORM mode.

STORM Research

Use this skill when the user needs a grounded research brief rather than a single-pass answer. The goal is not to make the answer longer; it is to make the thinking harder to fool.

Operating Modes

Choose the smallest mode that fits the request:

• Quick source check: verify the main claim, cite the best sources, name uncertainty, and give a concise answer.
• STORM brief: run the full multi-perspective pass, contradiction map, synthesis, and peer review.
• Full research swarm: split the work into lanes or agents when tools/routing support it, then reconcile the outputs into one brief.

If the user or agent profile names a mode, use that mode. Otherwise use STORM brief for important or ambiguous research.

Workflow

1. Frame the question

   • Restate the decision or topic in one sentence.
   • Identify what would change the user's action or belief.
   • Note time sensitivity and browse/current-source needs.

2. Gather evidence

   • Use current sources when facts may have changed.
   • Prefer primary sources, official docs, papers, filings, reputable data, and dated reporting.
   • Track source title, URL or vault path, date, core claim, and limits.

3. Run expert lenses

   • Practitioner: what daily operators know that outsiders miss.
   • Academic: what peer-reviewed or systematic evidence says.
   • Skeptic: strongest counterargument and ignored evidence.
   • Incentives: who benefits, pays, profits, or has distorted incentives.
   • Historian: comparable past patterns and how they resolved.

4. Map contradictions

   • List specific claims that clash.
   • Identify which side has stronger evidence and why.
   • Name what every lens agrees on.
   • Name the blind spot none of the lenses handled.

5. Synthesize and peer-review

   • Rank findings by reliability.
   • Separate sourced facts, inference, and assumptions.
   • Give confidence scores or low/medium/high confidence.
   • Identify the weakest claim and what would verify it.
   • Call out overrepresented perspectives and missing lenses.

Output Contract

When producing a research brief, use these exact markdown headings so HivemindOS can render the brief as tabs:

## Perspectives
## Contradictions
## Synthesis
## Peer Review
## Sources

Keep each section proportional to the mode. Quick source checks can be short; full research swarm outputs can be denser.

Source Discipline

• Cite sources for claims that depend on external facts.
• Include dates for time-sensitive sources.
• Do not smooth over conflicting evidence.
• Label assumptions and inference explicitly.
• If browsing or source access fails, state what could not be verified and continue with the safest available framing.

Full Research Swarm Pattern

When tools or HivemindOS routing support multiple lanes, split work into:

• source scout: primary/current evidence
• practitioner lens: operational reality
• skeptic lens: counterevidence and failure modes
• incentives lens: money, power, and adoption pressure
• synthesis reviewer: contradiction map and final confidence

If no multi-agent routing is available, simulate these lanes in one pass and clearly label it as a solo synthesis.