Who it is forA researcher starting a new analysis who wants the same plot style and code as a past paper, not a redrawn approximation.
The problemAsking an agent to recreate a figure gives a similar plot with different code, so styles drift between papers.
The outcomeThe confirmed template's exact files copied into your project, ready to adapt to new data, with nothing copied until you approve it.
What you need first
- An MCP host such as Pi, DeepSeek Harness (dsh), Claude, Codex or Cursor
- Node/npm for the Pi install
Steps
- Open the new project in your MCP host with the figure library installed
- Ask the agent to open the workbench and search your Local Published templates
- Preview candidates and confirm one card
- Let the agent plan materialization into an empty folder you name
- Approve the plan so the exact files are written
- Swap in your new data inputs and run the code with your own R or Python setup
Prompt for your agent
Open the figure library workbench and search my Local Published templates for <plot type>. Wait for me to confirm one card. Then plan materialization into <empty folder>. Do not execute R or Python. Do not redraw the figure.
Caveat. Drawing new data still needs approved R or Python runtimes in your host.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for xuzhougeng/ScientificFigureLibrary and "Reuse an exact figure template in a new project"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer