DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to keep a versioned library of your figure templates with Scientific Figure Library

Install Scientific Figure Library as an MCP server, bind one global Library folder, and import figures with their code as reviewed, immutable releases.

Short answer: For Pi run pi install npm:pi-mcp-adapter and then pi install npm:scientific-figure-library; then For other stdio hosts, follow docs/QUICKSTART.md and register the server figure-library pointing at dist/index.js (Node.js 22 or newer); Restart the host and call figure_library_get_skill, then figure_library_source_status. All steps.

  • Top pick
  • Research
  • Intermediate
  • 30 minutes
  • From the README
  • Codex
  • Cursor
  • Claude Code

Who it is forA researcher or bioinformatician who reuses plotting code across papers and wants every figure template versioned with the exact code that made it.

The problemFigure scripts are scattered across old projects, and nobody remembers which version produced the plot in the published paper.

The outcomeOne local library of reviewed figure templates, each release hashed and versioned, searchable from Claude, Codex, Cursor, Pi or dsh.

What you need first

  • An MCP host such as Pi, DeepSeek Harness (dsh), Claude, Codex or Cursor
  • Node/npm for the Pi install

Steps

  1. For Pi run pi install npm:pi-mcp-adapter and then pi install npm:scientific-figure-library
  2. For other stdio hosts, follow docs/QUICKSTART.md and register the server figure-library pointing at dist/index.js (Node.js 22 or newer)
  3. Restart the host and call figure_library_get_skill, then figure_library_source_status
  4. If it reports setup_required, bind one absolute global Library directory and a Local workspace
  5. Import a figure image with its code and review it at the gate
  6. Publish an immutable Release so it shows up under Local Published

Prompt for your agent

Call figure_library_source_status. If writes are disabled, help me bind <absolute path> as my global Library after I confirm. Then import the figure <image path> with its code <script path>, review it and publish a release named <name>. Do not execute R or Python.

Caveat. The server does not run plotting code; your host agent inspects and runs files.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for xuzhougeng/ScientificFigureLibrary and "Keep a versioned library of your figure templates"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer