Agent skill

CREATE_DEEP_RESEARCH_PROMPT

Generate a structured deep research prompt. USE WHEN the user needs to research a complex topic, market, or bug deeply. Capabilities include objective clarification, tech stack validation, and prompt generation.

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/create-deep-research-prompt

SKILL.md

CREATE DEEP RESEARCH PROMPT

Owner: Manny (Product Manager)

Goal

To translate a vague user request (e.g., "Research AI frameworks") into a highly specific, effective prompt for a Deep Research tool or agent.

Core Principles

  • Context is King: A research prompt without clear objectives yields noise.
  • Tech Validation: If researching technology, use context7 to ensure terms and versions are correct.
  • Classification: Different research (Market vs. Bug vs. Tech) requires different prompt structures.

Workflow Instructions

1. Classification

  • Action: Determine the research type.
    • Product Validation (Do users actually need this?)
    • Technical Review (Is this feasible?)
    • Bug Diagnosis (Why is this breaking?)
    • Market Analysis (Who are the benchmarks?)

2. Objective Clarification

  • Action: Elicit the "Why."
  • Prompt: "What decision will this research enable?" or "What is the single most important question to answer?"
  • Constraint: Stop if the goal is "just curious." Push for actionable intent.

3. Context & Tech Validation

  • Action: If the research involves specific tools/libraries:
    • Call context7 MCP to verify the library names and current versions.
    • Example: If user says "Research Next.js router," verify if they mean Pages or App Router.

4. Constructing the Prompt

  • Action: Assemble the research directive.
  • Structure:
    • Objective: The "North Star" question.
    • Key Questions: 3-5 specific sub-questions.
    • Methodology: How to search (e.g., "Compare documentation," "Search forums," "Read whitepapers").
    • Sources: Define authorized sources (e.g., "Official docs only," "GitHub issues," "Reddit").
    • Deliverable: What the output should look like (Table, Summary, Code snippets).

5. Review & Export

  • Action: Present the draft prompt to the user.
  • Refine: Incorporate feedback.
  • Output: Save the final prompt as a Markdown file in outputs/research_prompts/.

Anti-Patterns

  • NEVER generate a generic "Tell me about X" prompt.
  • NEVER include hallucinated library versions in a research prompt.
  • NEVER skip the "Success Criteria" definition.

Output Format

A Markdown file containing the structured prompt, ready to be fed into a reasoning model or research agent.

Didn't find tool you were looking for?

Be as detailed as possible for better results