AgentSkillsCN

Pdf Finder

PDF查找器

SKILL.md

PDF Finder

Find and download PDF papers from multiple open access sources.

Definition

  • ID: pdf-finder
  • Type: Command
  • Author: Research Skills
  • Description: Locate and download papers from Zotero, Unpaywall, Semantic Scholar, and arXiv

Commands

Find by DOI

code
/pdf-finder "10.1145/3706598.3714033"

Find PDF source for a paper by DOI.

Download

code
/pdf-finder --doi "10.1145/3706598.3714033" --download /path/to/papers

Download PDF to specified directory.

Find by Zotero Key

code
/pdf-finder --key ZOTERO_KEY

Find PDF for an item already in your Zotero library.

Batch Processing

code
/pdf-finder --batch dois.txt --download /path/to/papers

Process multiple DOIs from a file.

Requirements

  • Python 3.10+
  • requests>=2.31.0, httpx>=0.25.0
  • Optional: pymupdf>=1.24.0 for PDF parsing
  • Optional: Zotero desktop app for local lookups

Setup

  1. Install as part of skills-research:

    bash
    cd ~/.claude/skills/skills-research
    
  2. Configure credentials:

    bash
    cd literature-search/pdf-finder
    cp .env.example .env
    nano .env  # Set UNPAYWALL_EMAIL
    

Environment Variables

  • UNPAYWALL_EMAIL - Email for Unpaywall API (any valid email works)
  • ZOTERO_USER_ID, ZOTERO_API_KEY - Zotero credentials (inherited if set)

Sources (Priority Order)

  1. Zotero (local library if Zotero app running)
  2. Unpaywall (open access via institutional repositories)
  3. Semantic Scholar (author preprints)
  4. arXiv (preprint server)

Examples

code
/pdf-finder "10.1145/3706598"
/pdf-finder "10.1145/3706598" --download ~/Papers
/pdf-finder --key ABC123DEF --json
/pdf-finder --batch papers_to_find.txt --download ~/Papers
/pdf-finder "10.1145/3706598" --parse-after --output-dir ~/Papers/sections

Notes

  • Respects rate limiting of all APIs
  • Fails gracefully if PDF not found (returns source information)
  • Can parse downloaded PDFs into per-section markdown files