literature-review

科研文献综述技能。当用户提到"综述"、"survey"、"find papers"、 "文献调研"、"文献检索"时自动触发。支持多数据源检索和结果综合。

SKILL.md

GitHub

Literature Review Skill

Overview

This skill helps researchers conduct comprehensive literature reviews by searching multiple academic databases and synthesizing findings.

Trigger Conditions

  • User mentions "文献综述", "综述", "survey"
  • Tasks involving academic paper search and analysis
  • Requests for research background investigation

Workflow

Step 1: Query Understanding

Analyze the user's research topic and extract key concepts:

  • Main research area
  • Specific focus or questions
  • Time range preferences
  • Language preferences

Step 2: Multi-Source Search

Execute searches across multiple databases:

  • ArXiv (preprints, latest research)
  • Semantic Scholar (comprehensive coverage)

Script: scripts/search_papers.py

Step 3: Result Synthesis

Aggregate and analyze search results:

  • Deduplicate papers across sources
  • Rank by relevance and citations
  • Identify key themes and gaps

See: SEARCH_STRATEGIES.md for advanced techniques See: SYNTHESIS_GUIDE.md for synthesis methodology

Output Format

{
  "query": "original search query",
  "papers": [
    {
      "title": "Paper Title",
      "authors": ["Author 1", "Author 2"],
      "year": 2024,
      "summary": "Brief abstract",
      "url": "https://...",
      "relevance": "high"
    }
  ],
  "synthesis": {
    "themes": ["Theme 1", "Theme 2"],
    "gaps": ["Gap 1"],
    "recommendations": ["Recommendation 1"]
  }
}

Best Practices

  1. Start with broad searches, then narrow down
  2. Cross-reference findings across sources
  3. Prioritize recent, highly-cited papers
  4. Look for review articles first for overview