SuperAPI AI Directory
Partner中文AI newsFavorites
Skillsresearch-workflow

research-workflow

Writing & research

View repository

Updated 10-16 07:27

About this skill

Research Workflow

Conduct thorough research on any topic and deliver a professional PDF research report with citations. This workflow connects search and research skills to go from a user's question to a polished, shareable document.

Sub-Skills

  • rebyteai/internet-search — Quick web search for facts, recent news, and specific data points
  • rebyteai/deep-research — Comprehensive multi-source research with citations (10+ sources, trend analysis, conflicting viewpoints)
  • anthropics/pdf — Generate the final research report as a professional PDF document

Workflow

Step 1: Understand the Research Question

Parse what the user wants to know. Identify:

  • Core question — What is the central topic or question to answer?
  • Scope — How broad or narrow? A quick overview or exhaustive analysis?
  • Audience — Technical experts, executives, general public?
  • Deliverable — What should the final report contain? (analysis, recommendations, data, comparisons)

If the request is ambiguous, ask clarifying questions. If it's clear, proceed directly.

Step 2: Initial Discovery

Use internet-search to get oriented:

  • Run 2-3 searches to understand the landscape
  • Identify key players, terms, and recent developments
  • Find authoritative sources to dive deeper into

This step is quick — just enough to know what you're dealing with.

Step 3: Deep Research

Use deep-research for comprehensive analysis:

  • Gather information from 10+ sources
  • Cross-reference facts across sources
  • Identify consensus views and conflicting opinions
  • Collect data, statistics, and expert quotes
  • Track all sources for citations

Let deep-research handle the heavy lifting. It will return structured findings with citations.

Step 4: Synthesize and Structure

Before writing the report, organize findings:

  1. Key findings — What are the 3-5 most important takeaways?
  2. Supporting evidence — What data and sources back each finding?
  3. Conflicting views — Where do sources disagree? Why?
  4. Gaps — What couldn't be determined? What needs more research?
  5. Outline — Structure the report logically:
    • Executive Summary
    • Background/Context
    • Key Findings (with subheadings)
    • Analysis/Discussion
    • Conclusions/Recommendations
    • Sources/References

Step 5: Generate PDF Report

Use anthropics/pdf to create the final document:

  1. Title page — Clear title, date, topic summary
  2. Executive summary — 1 paragraph with key takeaways (for busy readers)
  3. Body sections — Each major finding as a section with evidence
  4. Citations — Inline references and a full bibliography
  5. Professional formatting — Headers, bullet points, tables where appropriate

The PDF should be self-contained — a reader should understand the topic without external context.

Step 6: Present Results

After generating the PDF:

  1. Share the PDF link with the user
  2. Provide a brief verbal summary:
    • Main conclusions
    • Most surprising finding
    • Any caveats or limitations
  3. Ask if the user wants:
    • Additional depth on any section
    • Different angle or focus
    • Follow-up research on related topics

Decision Points

  • "Should I use internet-search or deep-research?" — Use internet-search for quick fact-finding and orientation. Use deep-research when you need comprehensive analysis from multiple sources. For most research requests, you'll use both: internet-search first to orient, then deep-research for depth.

  • "How many sources is enough?" — For a credible research report, aim for 10+ sources minimum. More complex topics may need 20+. Quality matters more than quantity — prefer authoritative sources.

  • "The sources conflict" — This is valuable information. Report the disagreement, explain why sources might differ (methodology, timing, bias), and present the most credible interpretation.

  • "I can't find enough information" — Be honest. Report what you found, clearly state what's missing, and explain why (topic too new, proprietary information, etc.). A report that acknowledges gaps is more valuable than one that pretends to be complete.

  • "The topic is too broad" — Ask the user to narrow it, or propose a focused angle yourself. "AI in healthcare" is too broad; "AI diagnostic tools for radiology in 2024-2025" is researchable.