Plotnine MCP Server
Enables creation of publication-quality statistical graphics using plotnine's grammar of graphics through natural language, supporting 20+ geometry types, multi-layer plots, and flexible theming for data visualization.
README
Plotnine MCP Server
A Model Context Protocol (MCP) server that brings ggplot2's grammar of graphics to Python through plotnine, enabling AI-powered data visualization via natural language.
Create publication-quality statistical graphics through chat using plotnine's Python implementation of R's beloved ggplot2. This modular MCP server allows Claude and other AI assistants to generate highly customizable visualizations by composing layers through the grammar of graphics paradigm.
Features
- 🎨 Multi-Layer Plots: Combine multiple geometries in a single plot (scatter + trend lines, boxplots + jitter, etc.)
- Multiple Data Sources: Load data from files (CSV, JSON, Parquet, Excel), URLs, or inline JSON
- Grammar of Graphics: Compose plots using aesthetics, geometries, scales, themes, facets, and coordinates
- 20+ Geometry Types: Points, lines, bars, histograms, boxplots, violins, and more
- Flexible Theming: Built-in themes with extensive customization options
- Statistical Transformations: Add smoothing, binning, density estimation, and summaries
- Faceting: Split plots by categorical variables using wrap or grid layouts
- Multiple Output Formats: PNG, PDF, SVG with configurable dimensions and DPI
Installation
1. Clone or download this repository
cd plotnine-mcp
2. Install dependencies
Using pip:
pip install -e .
For full functionality (parquet and Excel support):
pip install -e ".[full]"
3. Configure Your MCP Client
Claude Desktop
Add the server to your Claude Desktop configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"plotnine": {
"command": "python",
"args": ["-m", "plotnine_mcp.server"]
}
}
}
If you installed in a virtual environment, use the full path to python:
{
"mcpServers": {
"plotnine": {
"command": "/path/to/venv/bin/python",
"args": ["-m", "plotnine_mcp.server"]
}
}
}
Cursor
Add to your Cursor settings by opening the command palette (Cmd/Ctrl+Shift+P) and searching for "Preferences: Open User Settings (JSON)". Add:
{
"mcp.servers": {
"plotnine": {
"command": "python",
"args": ["-m", "plotnine_mcp.server"]
}
}
}
Or configure via .cursor/mcp.json in your project:
{
"mcpServers": {
"plotnine": {
"command": "python",
"args": ["-m", "plotnine_mcp.server"]
}
}
}
VSCode (with Cline/Roo-Cline)
Add to your VSCode MCP settings file:
macOS/Linux: ~/.config/Code/User/globalStorage/rooveterinaryinc.roo-cline/settings/cline_mcp_settings.json
Windows: %APPDATA%\Code\User\globalStorage\rooveterinaryinc.roo-cline\settings\cline_mcp_settings.json
{
"mcpServers": {
"plotnine": {
"command": "python",
"args": ["-m", "plotnine_mcp.server"]
}
}
}
For other MCP clients in VSCode, consult their specific documentation for MCP server configuration.
4. Restart Your Application
Restart Claude Desktop, Cursor, or VSCode for the changes to take effect. The plotnine MCP server should now be available!
Usage
Basic Example
Create a scatter plot from data.csv with x=age and y=height
Advanced Example
Create a line plot from sales_data.csv showing:
- x: date, y: revenue, color by region
- Use a minimal theme with figure size 12x6
- Add a smooth trend line
- Facet by product category
- Label the plot "Q4 Sales Performance"
- Save as PDF
Available Tools
create_plot
Create a plotnine visualization with full customization.
Required Parameters:
-
data_source: Data source configurationtype: "file", "url", or "inline"path: File path or URL (for file/url types)data: Array of objects (for inline type)format: "csv", "json", "parquet", or "excel" (auto-detected)
-
aes: Aesthetic mappings (column names)x,y: Axis variablescolor,fill: Color aestheticssize,alpha,shape,linetype: Additional aestheticsgroup: Grouping variable
-
geom: Geometry specificationtype: Geometry type (point, line, bar, etc.)params: Additional parameters (size, alpha, color, etc.)
Optional Parameters:
scales: Array of scale configurationstheme: Theme configuration with base and customizationsfacets: Faceting configurationlabels: Plot labels (title, x, y, caption, subtitle)coords: Coordinate system configurationstats: Statistical transformationsoutput: Output configuration (format, size, DPI, directory)
list_geom_types
List all available geometry types with descriptions.
Geometry Types
- point: Scatter plot points
- line: Line plot connecting points
- bar: Bar chart (counts by default)
- col: Column chart (identity stat)
- histogram: Histogram of continuous data
- boxplot: Box and whisker plot
- violin: Violin plot for distributions
- area: Filled area under line
- density: Kernel density plot
- smooth: Smoothed conditional means
- jitter: Jittered points (reduces overplotting)
- tile: Heatmap/tile plot
- text: Text annotations
- errorbar: Error bars
- hline/vline/abline: Reference lines
- path: Path connecting points in order
- polygon: Filled polygon
- ribbon: Ribbon for intervals
Examples
Simple Scatter Plot
{
"data_source": {
"type": "file",
"path": "./data/iris.csv"
},
"aes": {
"x": "sepal_length",
"y": "sepal_width",
"color": "species"
},
"geom": {
"type": "point",
"params": {"size": 3, "alpha": 0.7}
}
}
Line Plot with Theme
{
"data_source": {
"type": "url",
"path": "https://example.com/timeseries.csv"
},
"aes": {
"x": "date",
"y": "value",
"color": "category"
},
"geom": {
"type": "line",
"params": {"size": 1.5}
},
"scales": [
{
"aesthetic": "x",
"type": "datetime",
"params": {"date_breaks": "1 month"}
}
],
"theme": {
"base": "minimal",
"customizations": {
"figure_size": [12, 6],
"legend_position": "bottom"
}
},
"labels": {
"title": "Time Series Analysis",
"x": "Date",
"y": "Value"
}
}
Faceted Boxplot
{
"data_source": {
"type": "inline",
"data": [
{"group": "A", "category": "X", "value": 10},
{"group": "A", "category": "Y", "value": 15},
{"group": "B", "category": "X", "value": 12}
]
},
"aes": {
"x": "group",
"y": "value",
"fill": "group"
},
"geom": {
"type": "boxplot"
},
"facets": {
"type": "wrap",
"facets": "~ category"
},
"theme": {
"base": "bw"
}
}
Multi-Layer Plot: Scatter + Smooth Trend
NEW! Layer multiple geometries to create complex visualizations:
{
"data_source": {
"type": "file",
"path": "./data/measurements.csv"
},
"aes": {
"x": "time",
"y": "value",
"color": "sensor"
},
"geoms": [
{
"type": "point",
"params": {"size": 2, "alpha": 0.6}
},
{
"type": "smooth",
"params": {"method": "lm", "se": false}
}
],
"theme": {
"base": "minimal",
"customizations": {"figure_size": [12, 6]}
},
"labels": {
"title": "Sensor Readings with Trend Lines",
"x": "Time",
"y": "Measurement"
}
}
Boxplot with Jittered Points
Show both distribution summary and individual data points:
{
"data_source": {
"type": "file",
"path": "./data/experiment.csv"
},
"aes": {
"x": "treatment",
"y": "response",
"fill": "treatment"
},
"geoms": [
{
"type": "boxplot",
"params": {"alpha": 0.7}
},
{
"type": "jitter",
"params": {"width": 0.2, "alpha": 0.5, "size": 1}
}
],
"theme": {
"base": "bw"
},
"labels": {
"title": "Treatment Effects with Individual Observations"
}
}
Chat Examples
You can create plots through natural language:
"Create a histogram of the 'age' column from users.csv"
"Make a scatter plot with smooth trend line showing price vs size, colored by category"
"Plot a line chart from sales.csv with date on x-axis and revenue on y-axis, faceted by region, using a dark theme"
"Create a violin plot comparing distributions of test scores across different schools"
"Make a boxplot with individual points overlaid showing temperature by season"
"Create a scatter plot with a linear trend line for each category, showing the relationship between hours studied and test scores"
Configuration Options
Themes
Available base themes:
gray(default)bw(black and white)minimalclassicdarklightvoid
Scale Types
- Positional: continuous, discrete, log10, sqrt, datetime
- Color/Fill: gradient, discrete, brewer
Coordinate Systems
cartesian(default)flip(swap x and y)fixed(fixed aspect ratio)trans(transformed coordinates)
Output
By default, plots are saved to ./output directory as PNG files with 300 DPI. You can customize:
- format: png, pdf, svg
- filename: Custom filename (auto-generated by default)
- width/height: Dimensions in inches
- dpi: Resolution for raster formats
- directory: Output directory path
Troubleshooting
"Module not found" errors
Ensure you've installed the package:
pip install -e .
Parquet/Excel support
Install optional dependencies:
pip install -e ".[full]"
"Cannot find data file"
Use absolute paths or paths relative to where Claude Desktop is running.
Plot not rendering
Check that:
- Column names in
aesmatch your data - Data types are appropriate for the geometry
- Required aesthetics are provided (e.g.,
xandyfor most geoms)
Development
Running tests
pytest
Code formatting
black src/
ruff check src/
License
MIT
Contributing
Contributions welcome! Please open an issue or submit a pull request.
Resources
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.