arxiv-search（arxiv 搜索）

arxiv-search: Academic Paper Search

Search the arXiv preprint repository for scholarly articles across physics, mathematics, computer science, quantitative biology, quantitative finance, statistics, and economics.

Quick Reference

# Basic search (auto-selects best implementation)
${CLAUDE_PLUGIN_ROOT}/arxiv_search "transformer attention mechanism"

# Limit results
${CLAUDE_PLUGIN_ROOT}/arxiv_search "protein folding" --max-papers 5

Implementation

The skill auto-detects the best available implementation:

1. Python (preferred): If uv + arxiv package or python3 + arxiv are available
2. Bash (fallback): Uses curl + awk - works everywhere, no dependencies

To enable Python implementation:

uv pip install arxiv

When to Use

• Finding recent research and preprints before journal publication
• Searching for ML/AI papers (cs.LG, cs.AI, cs.CV, stat.ML)
• Looking up mathematical or statistical methods
• Finding quantitative biology and bioinformatics papers (q-bio)
• Accessing physics and mathematics literature

Arguments

Output Format

Returns formatted results with:

• Title: Paper title
• Summary: Abstract text

Papers are separated by blank lines for readability.

Examples

Machine learning research:

${CLAUDE_PLUGIN_ROOT}/arxiv_search "large language models reasoning" --max-papers 5

Physics papers:

${CLAUDE_PLUGIN_ROOT}/arxiv_search "quantum computing error correction"

Statistics and methods:

${CLAUDE_PLUGIN_ROOT}/arxiv_search "bayesian inference neural networks"

arXiv Categories

Common categories for reference:

• cs.LG - Machine Learning
• cs.AI - Artificial Intelligence
• cs.CV - Computer Vision
• stat.ML - Statistics: Machine Learning
• q-bio - Quantitative Biology
• math - Mathematics
• physics - Physics
• q-fin - Quantitative Finance

Notes

• Papers are preprints and may not be peer-reviewed
• Results sorted by relevance to query
• No API key required - free access
• Python implementation is more robust for edge cases
• Bash fallback works on any Unix system with curl