arxiv-download-paper

name	arxiv-download-paper
description	Download academic papers from arXiv.org in PDF format. Use when user asks to download research papers, save academic papers locally, or get PDF versions of arXiv papers. Supports downloading by arXiv ID, title search, author search, or keyword query. Papers are saved with sanitized titles as filenames to the downloads directory. Optionally adds papers to knowledge base.
version	1.0.0
author	miniClaw
tags	["arxiv","download","pdf","academic","research","papers"]
dependencies	{"python":["arxiv>=2.0.0"]}

arXiv Paper Download

Download academic papers from arXiv.org in PDF format with proper filename sanitization.

Overview

This skill enables downloading PDF versions of academic papers from arXiv.org. Papers are saved with sanitized titles as filenames to prevent filesystem issues. Supports multiple search methods and optional knowledge base integration.

Quick Start

Download by arXiv ID

Download a specific paper using its arXiv ID:

python data/skills/arxiv-download-paper/scripts/download_paper.py --id "2307.09288" --output-dir downloads

Download by Title Search

Search and download papers by title:

python data/skills/arxiv-download-paper/scripts/download_paper.py --title "Attention Is All You Need" --output-dir downloads

Download Multiple Papers

Download multiple papers by keyword query:

python data/skills/arxiv-download-paper/scripts/download_paper.py --query "transformers" --max-results 5 --output-dir downloads

Download Parameters

The download_paper script accepts the following parameters:

Parameter	Short	Description	Example
--id	-i	Download specific paper by arXiv ID	"2307.09288"
--title	-t	Search by paper title	"Attention Is All You Need"
--query	-q	Search query (keywords, phrases)	"large language models"
--author	-a	Search by author name	"Geoffrey Hinton"
--output-dir	-o	Output directory for PDFs (default: downloads)	"downloads"
--max-results	-m	Maximum number of results (default: 10)	20
--sort-by	-s	Sort order: relevance, lastUpdatedDate, submittedDate	relevance
--order	-o	Sort direction: ascending, descending	descending
--category	-c	Filter by arXiv category	cs.AI
--add-to-kb		Add downloaded papers to knowledge base

Filename Sanitization

Papers are saved with filenames derived from their titles. Invalid characters are replaced with underscores:

• Invalid characters: < > : " / \ | ? *
• Control characters are removed
• Filenames are trimmed and limited to 200 characters
• Example: "Attention Is All You Need" → Attention_Is_All_You_Need.pdf

Common Workflows

Workflow 1: Download Single Paper

Download a specific paper by arXiv ID:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --id "2307.09288" \
  --output-dir downloads

Workflow 2: Download Papers by Topic

Download recent papers on a specific topic:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --query "diffusion models" \
  --category cs.LG \
  --sort-by submittedDate \
  --order descending \
  --max-results 10 \
  --output-dir downloads

Workflow 3: Download Papers by Author

Download papers from a specific researcher:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --author "Yann LeCun" \
  --max-results 5 \
  --output-dir downloads

Workflow 4: Download and Add to Knowledge Base

Download papers and add them to the knowledge base for future reference:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --id "2307.09288" \
  --output-dir downloads \
  --add-to-kb

Output Structure

Downloaded papers are saved to the specified output directory (default: downloads/):

downloads/
├── Attention_Is_All_You_Need.pdf
├── BERT_Pre-training_of_Deep_Bidirectional_Transformers.pdf
└── ...

If --add-to-kb is used, papers are also copied to knowledge_base/papers/.

Error Handling

• No results found: The query may be too specific. Try broader terms.
• Invalid arXiv ID: Check the ID format (e.g., 2307.09288).
• Download failures: Network issues or invalid PDF URLs. Check internet connection.
• File already exists: Script skips existing files to avoid overwrites.

Tips for Effective Downloads

1. Use specific IDs: When you know the exact paper, use --id for fastest download.
2. Limit results: Use --max-results to control how many papers to download.
3. Check categories: Add --category cs.AI to narrow results to relevant fields.
4. Sort by date: Use --sort-by submittedDate --order descending for latest papers.
5. Use knowledge base: Add --add-to-kb to build a searchable paper collection.

Example Interactions

User: "Download the paper 'Attention Is All You Need'"

Agent action:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --title "Attention Is All You Need" \
  --output-dir downloads

User: "Download arXiv paper 2307.09288"

Agent action:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --id "2307.09288" \
  --output-dir downloads

User: "Download 5 recent papers about transformers and add to knowledge base"

Agent action:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --query "transformers" \
  --sort-by submittedDate \
  --order descending \
  --max-results 5 \
  --output-dir downloads \
  --add-to-kb

User: "Download papers by Geoffrey Hinton"

Agent action:

python data/skills/arxiv-download-paper/scripts/download_paper.py \
  --author "Geoffrey Hinton" \
  --max-results 10 \
  --output-dir downloads

Notes

• Requires the arxiv Python package. Install with: pip install arxiv
• Downloads use urllib for better compatibility across platforms
• Progress is shown during download with percentage and bytes transferred
• Existing files are not overwritten to prevent data loss
• Knowledge base integration copies papers to knowledge_base/papers/ directory

Resources

scripts/download_paper.py

Python script that handles paper search and PDF download. Supports multiple search methods and filename sanitization.

Usage examples:

# Command line
python scripts/download_paper.py --id "2307.09288" --output-dir downloads

# With knowledge base
python scripts/download_paper.py --query "neural networks" --max-results 3 --add-to-kb