Research

Research

When to use

• A research direction or gap has been identified (typically after brainstorm)
• User needs a systematic search strategy for literature
• User wants to delegate literature discovery to Gemini Deep Research or GPT Deep Research
• User prefers to search databases manually and needs a structured keyword kit

CRITICAL: This skill does NOT do research itself. It generates search strategy and prompts. Actual literature discovery is delegated to Deep Research agents or done manually by the user.

Procedure

Step 1 — Clarify the search target

If not already clear, ask:

• What is the specific research question or gap you are searching around?
• What field and subfield?
• Do you have any anchor papers or authors already? (These help seed the search.)
• Are you looking for theoretical foundations, empirical studies, or both?
• Language preference: do you need Chinese-language literature (CNKI, CSSCI)?

Step 2 — Generate both paths

By default, generate both Path A and Path B in one response. Present both paths simultaneously — the user may want to delegate some searches (Path A) while doing others manually (Path B). Do not require the user to choose first.

• Path A: Delegated Deep Research — ready-to-copy prompt for Gemini or GPT Deep Research. Faster, broader synthesis, good first pass.
• Path B: Manual Search Kit — keywords and database recommendations the user searches themselves. More control, better for niche topics.

Step 3A — Path A: Delegated Deep Research

Generate a ready-to-copy prompt the user can paste directly into Gemini Deep Research or GPT Deep Research:

Topic: [specific research question]
Field: [user's field and subfield]
Find: peer-reviewed journal articles, highly cited
For each paper return:
- Authors, year, journal
- Citation count
- Abstract
- Main argument/finding
- Why it's relevant to [specific research question]
- Citekey suggestion (author+year format)

Prioritize: [specific journals or authors if known]
Keywords (EN): [list of English search terms]
Keywords (ZH): [list of Chinese search terms]

Tell the user: "Paste this into Gemini Deep Research or GPT Deep Research. When you have results, run ingest to process them into the project."

Step 3B — Path B: Manual Search Kit

Generate a structured search kit:

Suggested keywords:
  English: [terms with Boolean combinations, e.g., "political polarization" AND "social media" NOT "echo chamber"]
  Chinese: [terms, e.g., 政治极化、社交媒体、舆论分化]

Suggested databases:
  CSSCI — Chinese Social Sciences Citation Index (Chinese journals)
  Web of Science — peer-reviewed international journals
  JSTOR — humanities and social sciences archive
  Google Scholar — broad discovery, check citations
  CNKI — Chinese National Knowledge Infrastructure (Chinese theses, journals)

Suggested authors to check: [names active in this area]
Suggested journals: [top journals for this topic and field]

Filter criteria:
  peer-reviewed only
  citation count > 50 for international journals, > 10 for Chinese journals (sociology defaults — adjust for your field)
  publication year: [suggest range based on topic]

Warning: Only suggest authors you are confident exist in this field. If you're not sure who the active researchers are, leave this field with a note like '[consult advisor for key authors in this subfield]' rather than guessing. Do NOT fabricate author names.

Tell the user: "When you have results, run ingest to process them into the project."

Inputs

• Research question or gap description
• Field and subfield
• Any existing anchor papers or authors
• Language requirements (English only, or Chinese + English)

Optional (if project exists):

• meta/memory.md — prior direction decisions and existing references
• references/literature/ — existing literature to avoid re-searching

Outputs

Path A: Ready-to-copy Deep Research prompt for Gemini or GPT Deep Research, including:

• Structured topic description
• Per-paper return format with citation count and citekey suggestion
• Keywords in English and Chinese

Path B: Manual search kit including:

• Boolean keyword combinations in English
• Chinese keyword list
• Recommended databases: CSSCI, Web of Science, JSTOR, Google Scholar, CNKI
• Suggested authors and journals
• Peer-reviewed and citation count filters

Both paths end at: "When you have results, run ingest to process them into the project."

Notes

• Never fabricate citations or pretend to search databases. This skill generates prompts and strategies only.
• Citekey format is author+year, e.g., bail2018, zhang2021 — consistent with Zotero BetterBibTeX convention used across the project.
• For Chinese social science topics, CSSCI and CNKI are essential — do not omit them.
• If the user already has some papers, ask for them before generating keywords — anchor papers dramatically improve keyword quality.
• After this skill, the next step is always ingest once the user has literature results.
• Path A (Gemini/GPT Deep Research) is recommended for initial broad sweeps. Path B is better for specialized or niche searches where automated tools underperform.