mcpField
F
figshare-data-download

figshare-data-download

github.com/tiangong-aiData & databases1140 installsUpdated 1w ago
Repository

About figshare-data-download

A skill for downloading Figshare dataset files through a browser when anti-bot checks are involved. It uses a browser-only flow to obtain the file, copy it to a target path, and verify the result.

Key features

  1. 01Open DOI or item pages in a browser
  2. 02Complete browser checks and start downloads
  3. 03Copy files to a specified path
  4. 04Verify file size and type

Use cases

  1. 01Download Figshare files behind anti-bot checks
  2. 02Save a downloaded file to a chosen local path

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/tiangong-ai/agent-skills --skill figshare-data-download

After installing, type/figshare-data-downloadin Claude Code to confirm it loaded.

Details

From the project README

Figshare Data Download

Use this skill to fetch Figshare dataset files behind anti-bot checks.

Workflow

  1. Normalize inputs:
  • item_or_file_url: DOI URL, item page URL, or ndownloader URL
  • output_path: final local path
  • expected_name: expected browser filename (optional)
  1. Resolve the file download link in browser:
  • Open the DOI/item page with browser tooling.
  • Find and click the dataset "Download" / "Download file" element.
  • If already given an ndownloader URL, open it directly in browser.

  1. Complete any browser challenge/verification and trigger the real file download.

  2. Copy the downloaded file from ~/Downloads to output_path.

  3. Verify output:

  • Ensure file size is non-zero.
  • Ensure extension/type matches expectation (.xlsx, .csv, etc.).

Decision Rules

  • Do not use web_fetch for this skill.
  • Do not run curl probing as part of this skill workflow.
  • Always use browser path first and keep it end-to-end.

Output Contract

  • Always produce one final file at output_path.
  • Keep the downloaded source file in ~/Downloads unchanged.
  • Report:
    • final path
    • file size
    • that the transfer used browser-only workflow

Script

  • scripts/figshare_data_download.py (browser-only helper)
    • supports semi-automatic flow: --open-browser + wait/poll in ~/Downloads + copy/verify
    • does not perform curl/web_fetch probing

Comments

Markdown supported