Academic Paper Fetcher

Fetch academic papers from Sci-Hub given a DOI. Automatically downloads PDFs and saves them to research/papers/ with clean filenames. Use when the user provides a DOI or requests a paper from PubMed.

Security Scan

The skill's code and instructions generally match the stated purpose (downloading PDFs from Sci‑Hub) but there are inconsistencies and undeclared requirements you should be aware of before installing.

Purpose & Capability

The skill's stated purpose (fetch papers by DOI from Sci‑Hub) matches the included script, which constructs Sci‑Hub URLs and downloads PDFs. However the SKILL metadata claims no required binaries while the script calls curl via subprocess; that binary is effectively required but not declared. SKILL.md also mentions 'use browser' and integrations (Obsidian, research automation) that are not implemented in the script—these are mismatches between description and actual capability.

Instruction Scope

SKILL.md instructs the agent to 'navigate' Sci‑Hub and save files to workspace/research/papers/ and to process multiple DOIs; the script implements that flow using curl and local file writes. The instructions do not ask to read unrelated system files or secrets. Minor scope creep: integration claims (Obsidian sync, automatic discovery from research runs) are described but there is no code to perform those integrations, so the agent would need additional logic to realize them.

Install Mechanism

There is no install spec (instruction-only skill with a bundled script). That keeps disk/write risk low. The included Python script will run as-is; there is no package download or archive extraction. Note: network access and curl must be available at runtime even though they are not declared.

Credentials

The skill declares no required environment variables or credentials, which is appropriate. However it fails to declare a required runtime dependency: the script invokes curl via subprocess.run. The mismatch between declared requirements (none) and actual runtime needs (curl, network access) is a proportionality/information issue that could cause unexpected failures or surprises.

Persistence & Privilege

The skill does not request permanent presence (always: false) and does not modify other skills or system-wide settings. It writes files only into the provided output directory (default current directory); the SKILL.md prescribes workspace/research/papers/ but the script will accept an output_dir parameter, which is reasonable behavior.

What to consider before installing

This skill mostly does what it says (downloads PDFs from a hard-coded Sci‑Hub domain). Before installing, consider: 1) Legal/ethical: Sci‑Hub distributes copyrighted material in many jurisdictions — ensure you are comfortable with that risk. 2) Undeclared runtime dependency: the Python script calls curl via subprocess but the skill metadata does not list curl as required; ensure curl and network access are available and acceptable. 3) Mismatched claims: SKILL.md mentions browser navigation and Obsidian/automation integrations that are not implemented in the script — don’t assume those features exist. 4) Safety: downloaded PDFs can contain exploits (rare) and files are written to your workspace path; run the script in a sandboxed environment if you are unsure. 5) If you want more confidence, ask the author to (a) declare required binaries (curl), (b) remove or implement claimed integrations, and (c) optionally allow selecting an alternative, lawful source (publisher APIs, institutional access) instead of Sci‑Hub.

Paper Fetcher

Automated academic paper retrieval from Sci-Hub.

Usage

Simple fetch:

Get paper: 10.1038/nature12345

Multiple papers:

Fetch these papers:
- 10.1016/j.cell.2023.01.001
- 10.1038/s41586-2023-06789-0
- 10.1126/science.abc1234

With context:

Get the epitalon paper: 10.1007/s12603-011-0032-7

What It Does

1. Takes DOI as input
2. Navigates to https://www.sci-hub.su/DOI
3. Downloads the PDF
4. Saves to research/papers/ with clean filename
5. Returns confirmation with file path

Output Location

workspace/
└── research/
    └── papers/
        ├── paper_10.1038_nature12345.pdf
        ├── paper_10.1016_j.cell.2023.01.001.pdf
        └── ...

Filename Format

paper_[DOI with slashes replaced].pdf

Examples:

• DOI: 10.1038/nature12345 → paper_10.1038_nature12345.pdf
• DOI: 10.1016/j.cell.2023.01.001 → paper_10.1016_j.cell.2023.01.001.pdf

Workflow

When user provides a DOI:

1. Extract DOI - Parse from message (with or without https://doi.org/ prefix)
2. Navigate Sci-Hub - Use browser to load https://www.sci-hub.su/DOI
3. Wait for PDF - Let page load and find download link
4. Download - Save PDF to research/papers/
5. Confirm - Report success with file path

Error Handling

If paper not found on Sci-Hub:

• Report that Sci-Hub couldn't find it
• Suggest checking the DOI format
• User can try manual search

If download fails:

• Report the error
• Provide Sci-Hub URL for manual download

Integration

With Obsidian Sync:

• Papers saved in research/papers/
• Can create notes linking to PDFs
• Sync metadata to Obsidian vault

With Research Automation:

• Fetch papers discovered in research runs
• Build reference library automatically
• Cross-reference with protocol notes

Tips

Finding DOIs:

• PubMed: Listed in article details
• Paper itself: Usually on first page
• Google Scholar: In citation info

Format flexibility:

• With prefix: https://doi.org/10.1038/nature12345 ✅
• Without prefix: 10.1038/nature12345 ✅
• Either format works

Batch fetching:

• Send multiple DOIs at once
• Processed sequentially
• All saved to research/papers/

---

Status: Active
Sci-Hub Domain: https://www.sci-hub.su
Save Location: research/papers/