paper-fetch — Legal Open-Access PDF Downloader

中文文档

What it does

• Downloads paper PDFs from a DOI (or batch file of DOIs) via legal open-access sources
• 5-source fallback chain: Unpaywall → Semantic Scholar openAccessPdf → arXiv → PubMed Central OA → bioRxiv/medRxiv
• Zero dependencies — pure Python standard library, no pip install needed
• Auto-named output — {first_author}_{year}_{short_title}.pdf
• Batch mode — pass a file of DOIs with --batch
• Never touches Sci-Hub or any paywall-bypass service — if no OA copy exists, reports failure with metadata so you can go through ILL

Discipline Coverage

The skill is discipline-agnostic — it works for any field, not just life sciences or computer science. Coverage depends on whether the paper has a legal OA version, not on its subject area.

In practice, Unpaywall + Semantic Scholar alone cover OA papers in chemistry, materials, economics, psychology, humanities, and every other field via institutional repositories, SSRN, RePEc, and publisher-hosted OA copies. arXiv/PMC/bioRxiv are additional fallbacks for their specific domains. If no legal OA copy exists anywhere, the skill reports failure honestly — it will never bypass paywalls regardless of discipline.

Multi-Platform Support

Works with all major AI coding agents that support the Agent Skills format:

Comparison

vs No Skill (native agent)

Prerequisites

• Python 3.8+ (standard library only, no extra packages)
• Unpaywall contact email (optional but recommended) — set once:

export [email protected]

Add it to ~/.zshrc / ~/.bashrc to persist. Without it, Unpaywall is skipped and the remaining 4 sources (Semantic Scholar, arXiv, PMC, bioRxiv/medRxiv) are still tried.

Skill Installation

Claude Code

# Global install
git clone https://github.com/Agents365-ai/paper-fetch.git ~/.claude/skills/paper-fetch

# Project-level install
git clone https://github.com/Agents365-ai/paper-fetch.git .claude/skills/paper-fetch

OpenClaw / ClawHub

clawhub install paper-fetch

# Or manual
git clone https://github.com/Agents365-ai/paper-fetch.git ~/.openclaw/skills/paper-fetch

Hermes Agent

git clone https://github.com/Agents365-ai/paper-fetch.git ~/.hermes/skills/research/paper-fetch

Or add to ~/.hermes/config.yaml:

skills:
  external_dirs:
    - ~/myskills/paper-fetch

pi-mono

git clone https://github.com/Agents365-ai/paper-fetch.git ~/.pimo/skills/paper-fetch

OpenAI Codex

# User-level
git clone https://github.com/Agents365-ai/paper-fetch.git ~/.agents/skills/paper-fetch

# Project-level
git clone https://github.com/Agents365-ai/paper-fetch.git .agents/skills/paper-fetch

SkillsMP

skills install paper-fetch

Installation paths summary

Usage

Single DOI:

python scripts/fetch.py 10.1038/s41586-021-03819-2

Custom output directory:

python scripts/fetch.py 10.1038/s41586-021-03819-2 --out ~/papers

Batch mode:

cat > dois.txt <<EOF
10.1038/s41586-021-03819-2
10.1126/science.abj8754
10.1101/2023.01.01.522400
EOF

python scripts/fetch.py --batch dois.txt --out ~/papers

Dry-run (preview without downloading):

python scripts/fetch.py 10.1038/s41586-020-2649-2 --dry-run

Human-readable text output:

python scripts/fetch.py 10.1038/s41586-020-2649-2 --format text

Or just ask your agent naturally:

Download the AlphaFold2 paper PDF to my ~/papers folder

Fetch the PDF for DOI 10.1038/s41586-020-2649-2

Download these three papers: 10.1038/s41586-021-03819-2, 10.1126/science.abj8754, 10.1101/2023.01.01.522400

Check if this paper has an open-access PDF available: 10.1038/s41586-020-2649-2

Batch download all DOIs from my dois.txt file into ~/papers

Resolution Order

1. Unpaywall — best OA location across all publishers (highest hit rate)
2. Semantic Scholar — openAccessPdf field + externalIds lookup
3. arXiv — if the paper has an arXiv ID
4. PubMed Central OA subset — if the paper has a PMCID
5. bioRxiv / medRxiv — DOI prefix 10.1101/
6. Otherwise → report failure with metadata (title/authors) for ILL

Files

• SKILL.md — the only required file. Loaded by all platforms.
• scripts/fetch.py — the downloader (pure stdlib Python)
• agents/openai.yaml — OpenAI Codex sidecar configuration
• README.md — this file
• README_CN.md — Chinese documentation

Known Limitations

• Coverage depends on OA availability — if a paper has no legal OA copy, this skill cannot get it. That is a feature, not a bug.
• Some publisher redirects return an HTML landing page instead of a PDF; the script validates the %PDF header and fails cleanly in that case
• No authentication — institutional proxies (EZproxy / OpenAthens) are not supported in this version
• Host allowlist — downloads are restricted to known OA provider domains; PDFs from unlisted hosts are blocked
• 50 MB size limit — per-PDF download cap to prevent runaway downloads

License

MIT

Support

If this skill helps your work, consider supporting the author:

WeChat Pay

Alipay

Buy Me a Coffee

Author

Agents365-ai

• Bilibili: https://space.bilibili.com/441831884
• GitHub: https://github.com/Agents365-ai