research-prompt_skill

Overview

This skill turns vague research requests into precise, high-impact prompts for deep research agents. It conducts a focused interview to gather decision context, audience, scope, and evidence preferences, then produces a concise, copy-pasteable research prompt tailored to the user’s needs. The goal is to save time and get actionable, verifiable results on the first run.

How this skill works

I act as an adaptive interviewer: I read the user’s initial message, ask 1–2 prioritized follow-up questions at a time, and stop once I have enough context to craft a strong prompt. The generated prompt embeds user constraints, defines what “good” looks like, requests structure (tables, comparisons, risk assessments) when useful, and adds anti-slop directives to avoid vague, padded answers. I deliver the prompt as a clean block plus a 1–2 sentence note explaining what I optimized for.

When to use it

• You need to research a topic but don’t know what to ask an AI research agent
• You want a precise prompt for a deep research tool (e.g., ChatGPT Deep Research)
• You must support a decision or action and need targeted, evidence-backed findings
• You have vague or partial context and want the prompt tailored to your knowledge level
• You need results organized for a specific audience or deliverable

Best practices

• Always state the decision or action the research supports; this shapes priorities
• Tell me what you already know so prompts skip basics or dive deeper as needed
• Share any time-sensitivity or source-type requirements up front
• Prefer short Q&A rounds: I’ll ask 1–2 follow-ups, then deliver the prompt
• Request desired output structure (comparison table, ranked options, executive summary) to make findings actionable
• Specify unacceptable outcomes (e.g., surface-level overviews) so I can include anti-slop instructions

Example use cases

• Deciding between technology vendors with a pros/cons, cost, and timeline comparison
• Preparing a board briefing on market trends with cited, recent sources
• Evaluating clinical evidence for a medical product with source-quality filters
• Creating a research prompt for competitive landscape and IP risks for investors
• Generating a structured investigation into regulatory changes and business impact