About this skill

Problem Solved

Deep Research prompts often scatter research scope, assumptions, and quality requirements across separate top-level sections. That can make a downstream AI research agent treat goals, source standards, and deliverables as disconnected tasks. This skill reframes a research request as an executable prompt: it does not answer the research question directly, but generates instructions for another AI research agent to run.

How It Works

The core capability is to rewrite a rough topic into a D1–D6 prompt. D1 Define captures the research goal, default assumptions, and boundaries. D2 Decompose identifies dimensions, regions, entities, or sub-questions. D3 Design Fields defines the fields to collect and what counts as complete. D4 Discover specifies search channels, source quality, and exclusion rules. D5 Digest requires deduplication, verification, conflict handling, and citation norms. D6 Deliver defines whether the final output is a table, report, or recommendation set.
The skill first infers the likely decision context, audience, and expected deliverable, then designs the search plan. If key details are missing, it makes conservative assumptions and places them in D1, asking only when the missing detail materially changes scope.

Boundaries and Notes

It is useful for structuring market analysis, product research, policy or legal research, academic research, and engineering or materials research into a controlled research task. It does not replace final fact-checking, and the generated content is a prompt rather than a research answer. Output defaults to Chinese unless another language is requested; for legal, standards, or experimental research, add verifiable sources, citation format, and uncertainty notes across D4–D6.

Screenshots

Use Cases

Market research: structure players, pricing, channels, customer groups, and trend evidence into a D1–D6 prompt.
Product research: define user scenarios, pain points, feature comparisons, and evidence confidence with clear delivery requirements.
Policy research: specify jurisdiction, effective dates, primary legal texts, enforcement bodies, and compliance implications.
Technical research: include definitions, methods, benchmark papers, datasets, evaluation metrics, and reproducibility constraints.

Best For

Industry analysts who need to turn competitor, channel, and trend material into a structured research prompt
Product managers who need to organize user scenarios, pain points, and feature comparisons with evidence
Legal researchers who need to map jurisdiction, effective dates, and compliance implications
Data scientists who need to verify methods, datasets, benchmark papers, and evaluation metrics