Scientific Figure Library MCP
Local MCP server and MCP App for scientific figures: import a figure with code, review it, publish a release, reuse the template
- 140 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Writes library files to the chosen disk and copies templates into projects on confirmation
- Checks a signed GitHub feed in the background and updates the local Open Figure Modules catalog
Install
- Works in
Manual install
npm run package:cursorBuilds the Cursor plugin and unzips into ~/.cursor/plugins/local/figure-library/. Node.js 22 or newer is required.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Writes library files to the chosen disk and copies templates into projects on confirmation
- Checks a signed GitHub feed in the background and updates the local Open Figure Modules catalog
This is third-party code. Review the repository files before installing.
What it does
Scientific Figure Library, or SFL, keeps your scientific figures in one global library on disk. You import a figure together with its code, review it and publish an immutable release. After that the agent searches for a matching template and shows an exact preview, and copies it into a project only after you confirm. The server does not execute plotting code and has no second model: the host agent reads the files while SFL hashes, versions, gates and publishes them. The default retrieval order is Local Published, then FigureYa, then Open Figure Modules and enabled personal providers. The frozen Community snapshot stays for compatibility and is excluded from default search. The tools come in plan and apply pairs: figure_library_plan_bind_global and apply_bind_global choose the library directory, figure_library_search searches, plan_working_revision and apply_working_revision import a figure and code unit, plan_publish_working_revision publishes a release and plan_materialize and apply_materialize copy a confirmed template into a project. figure_library_open opens the MCP App gallery. Backup, restore and forking of the library are available.
Who it is for. Researchers and bioinformaticians who build similar scientific figures across projects and want verified templates in one place.
Good fit when
- The same plot with its code has to be copied from project to project
- You need to import a figure with its code, review it and pin it as an immutable release
- Before copying into a project you want an exact template preview and a confirmation
- The library has to move to another machine: backup and restore
- Template search should go first through your own figures, then FigureYa and Open Figure Modules
Not a fit when
- You need to run plotting code: the server does not execute it
- You need a new figure generated by a model: there is no second model inside
- The environment has no Node.js 22 or newer: the server is built on it
- You need a shared cloud library: the data sits on a disk you choose
Example request
Find a volcano plot template in my library, show a preview and copy it into the current project after I confirmMCP
- Transport
- stdio
- Authentication
- not required
Limitations
- Node.js 22 or newer is required. For Pi and DeepSeek Harness install through the scientific-figure-library npm package, since a repository clone does not work: it has no dist/ folder.
- Before searching you must bind one global library directory and a local workspace; without them the server reports setup_required.
- The frozen Community snapshot is excluded from search by default, and an extra catalog can hold zero releases after an authorized redaction.
- The local client for macOS, Windows and Linux ships as a preview.
How to disable. Remove the figure-library entry from the host MCP settings and, for Pi, uninstall the npm package. The library folder on disk stays and is deleted manually.
FAQ
How do I connect Scientific Figure Library to Cursor or Codex?
Give the agent the repository and the README request: it follows docs/QUICKSTART.md and registers the stdio server figure-library at dist/index.js. For Cursor there is npm run package:cursor and unzip into ~/.cursor/plugins/local/figure-library/.
How do I install it in Pi?
Run pi install npm:pi-mcp-adapter and pi install npm:scientific-figure-library, then restart Pi. A clone or a GitHub URL install does not work.
Does the server run my plotting code?
No, the server never runs notebooks, installers or plot scripts. The agent reads the files while SFL hashes, versions and publishes releases.
Where are the figures stored?
In one global library directory on a disk you choose. A template is copied into a project only after confirmation.
What does setup_required mean?
The library is not bound yet. Bind a global directory and a local workspace, then repeat the search.
Discussion
No questions or reviews yet.
Related
MCP serversA skill that builds scientific figures from a table: it offers 2–3 chart candidates, assembles panels and checks image reuse.
Semiotic MCP
Semiotic
Semiotic's MCP server: an agent picks the chart type, renders SVG and validates the configuration against component schemas.
A visualization language for agents and an MCP server: neat charts from a simple semantic spec
An MCP server of science agents: Claude Code analyzes spectra and images, plans experiments and prepares DFT calculations
MCP server and CLI that turns paper text or data into academic figures, diagrams and plots
The official Graphlit MCP server: an agent collects material from Slack, GitHub, mail and websites into a knowledge base and searches it
README in short
The page presents SFL as a local MCP server and MCP App for scientific figures, explains the path from import to release and the search order, then gives agent-driven install for Pi, DeepSeek Harness and stdio MCP hosts, the feature set, a tools table and a description of Open Figure Modules.