inkscape_mcp
MCP server that lets AI agents drive Inkscape — interactively alongside the GUI or headlessly from the CLI
README
Inkscape MCP
<p align="center"> <img src="assets/social-preview.svg" alt="inkscape-mcp — drive Inkscape from an AI agent" width="100%"> </p>
Let your AI agent design, edit, and analyze vector graphics through Inkscape — live alongside a running Inkscape window, or headless from the command line.
<p> <a href="https://pypi.org/project/inkscape-mcp/"><img src="https://img.shields.io/pypi/v/inkscape-mcp.svg?style=flat-square&logo=pypi&logoColor=white&label=PyPI" alt="PyPI"></a> <a href="https://pypi.org/project/inkscape-mcp/"><img src="https://img.shields.io/pypi/pyversions/inkscape-mcp.svg?style=flat-square&logo=python&logoColor=white&label=Python" alt="Python versions"></a> <a href="https://pypi.org/project/inkscape-mcp/"><img src="https://img.shields.io/pypi/dm/inkscape-mcp.svg?style=flat-square&label=downloads" alt="PyPI downloads"></a> <a href="https://github.com/aravindev/inkscape_mcp/actions/workflows/ci.yml"><img src="https://img.shields.io/github/actions/workflow/status/aravindev/inkscape_mcp/ci.yml?branch=main&style=flat-square&logo=github&label=CI" alt="CI"></a> <a href="https://github.com/aravindev/inkscape_mcp/blob/main/LICENSE"><img src="https://img.shields.io/github/license/aravindev/inkscape_mcp?style=flat-square&color=yellow" alt="License"></a> </p>
<p> <a href="https://modelcontextprotocol.io"><img src="https://img.shields.io/badge/MCP-1.x-1F6FEB?style=flat-square&logo=anthropic&logoColor=white" alt="MCP"></a> <a href="https://github.com/PrefectHQ/fastmcp"><img src="https://img.shields.io/badge/FastMCP-3.2-7c5cfc?style=flat-square" alt="FastMCP"></a> <a href="https://inkscape.org"><img src="https://img.shields.io/badge/Inkscape-1.4.4-000000?style=flat-square&logo=inkscape&logoColor=white" alt="Inkscape"></a> <a href="https://github.com/astral-sh/ruff"><img src="https://img.shields.io/badge/style-ruff-D7FF64?style=flat-square&logo=ruff&logoColor=black" alt="Ruff"></a> <a href="https://github.com/pre-commit/pre-commit"><img src="https://img.shields.io/badge/pre--commit-enabled-success?style=flat-square&logo=pre-commit&logoColor=white" alt="pre-commit"></a> </p>
With Inkscape MCP, your coding agent can do everything a designer does in Inkscape — sketch and reshape paths, apply effects and filters, render LaTeX equations, generate barcodes and QR codes, convert between formats, query and rewrite document structure — through plain conversation instead of menu clicks.
Pair it with a running Inkscape window and the agent works alongside you — it sees what you have selected, opens property dialogs on your behalf, applies edits that appear on the canvas in real time (each one a normal undo entry you can roll back), and rasterizes its own work to inspect and refine the result. Or run it headlessly to batch-convert artwork, automate exports, and build pipelines around vector assets — no GUI required. The full Inkscape feature surface, including its hundreds of bundled extensions, is reachable as tool calls.
How it works
- The agent calls one of eight portmanteau tools —
inkscape_file,inkscape_vector,inkscape_analysis,inkscape_system,inkscape_extension,inkscape_gradient,inkscape_metadata, orinkscape_live— with anoperationstring that picks the actual behavior. - For headless work the server dispatches through
InkscapeCliWrapper, which builds and runs aninkscape --actions=...(or--export-...) command. For live-canvas work it talks to a running Inkscape window over D-Bus and stages SVG fragments through the clipboard. - Inkscape reads your input file (or the open document), runs the operation, and writes the output.
- The server returns dimensions, file paths, validation results, or whatever the operation produces — typed and trimmed for the agent's context window.
No external network. Just the Inkscape binary, an optional running Inkscape window, and your local files.
Getting started
Tested with Inkscape 1.4.4 on Ubuntu 24.04.
# 1. Install Inkscape + system headers for pycairo / pygobject
sudo add-apt-repository ppa:inkscape.dev/stable
sudo apt update
sudo apt install inkscape libcairo2-dev libgirepository-2.0-dev libgirepository1.0-dev \
pkg-config python3-dev xclip
# xclip stages SVG fragments for clipboard-based insert into the canvas.
# On Wayland substitute wl-clipboard for xclip — clipboard ops auto-detect.
# 2. Register with Claude Code (uvx fetches inkscape-mcp from PyPI on first run)
claude mcp add inkscape_mcp -s user -- uvx inkscape_mcp
claude mcp list | grep inkscape_mcp # → ✓ Connected
# 3. (Re)start Inkscape so it picks up the bundled extensions
# The MCP server auto-installs its inkex extensions to
# ~/.config/inkscape/extensions/inkscape_mcp/ on first boot. Inkscape only
# scans the extensions dir at startup, so close + reopen Inkscape once.
# Verify via:
# inkscape_system(operation="diagnostics") → bridge.needs_restart should be false
That's it. Ask the agent to convert an SVG, query an object's bounding box, run a boolean union — it'll pick the right tool.
Alternative install paths
# Plain pip into a venv
pip install inkscape-mcp
# uv into a project
uv add inkscape-mcp
The PyPI name is inkscape-mcp (hyphen). The Python import name is inkscape_mcp (underscore — PEP 8 / valid identifier). Both pip install inkscape-mcp and pip install inkscape_mcp resolve to the same project (PEP 503 name normalization).
Configuration
The Claude Code entry the install command above creates looks like:
{
"mcpServers": {
"inkscape_mcp": {
"command": "uvx",
"args": ["inkscape_mcp"]
}
}
}
Pin a non-default Inkscape with INKSCAPE_BIN=/path/to/inkscape in the server env. Inspect from the agent via inkscape_system(operation="diagnostics") — it reports which value came from which source (env var, config file, default).
Try asking your agent to…
Editing the file you have open
- "Recolor everything blue to red"
- "Open document properties — I can't find the DPI setting"
- "Select the pen tool so I can draw a curve"
- "Show me the XML editor"
- "Hide the snap controls bar"
- "Add a drop shadow to all the rectangles"
- "Convert this monogram to a single compound path"
- "Run a boolean union on the two selected shapes"
Generating SVG content
- "Drop in an isometric grid into this document"
- "Generate a calendar for the current month and paste it here"
- "Make me a 24×24 settings cog icon — currentColor stroke, round caps"
Inspection & format conversion
- "Export this to a 4× PNG and a print-ready PDF"
- "What's the bounding box of #logo-mark?"
- "Optimize this SVG — strip Inkscape namespace cruft"
- "List every object in the file with its type and size"
Live-canvas workflows (Inkscape running)
- "Read what I have selected"
- "Align these three icons horizontally with equal spacing"
- "Swap the whole palette to a warmer earth-tone scheme"
- "Add a 3-stop gradient to the rect with id
hero"
The agent picks the right tool. It can drive 1,073 of Inkscape's actions over D-Bus, fall back to headless CLI when the GUI isn't running, and run inkex extensions for things D-Bus doesn't cover.
Available tools
inkscape_file— load, save, convert (PDF/PNG/EPS/PS/…), validate, batch convert.inkscape_vector— 47 ops covering boolean, tracing, path simplification, optimization, layer export, geometry mutation, barcodes, LPE chains, tile clones.inkscape_analysis— quality scoring, statistics, object lists, layer structure.inkscape_system— server status, diagnostics, extension discovery and execution.inkscape_extension— discover and invoke any installed inkex extension (headless or on the live canvas).inkscape_gradient— gradient-stop manipulation; linear↔radial conversion.inkscape_metadata— read/write Dublin-Core RDF metadata.inkscape_live— drive the running GUI over D-Bus: live edits, paste-in-place, XML edits,rasterizeso the agent can see the canvas (preserves undo).
Full operation list in docs/TOOLS.md.
Learn more
- docs/INSTALL.md — full install walkthrough.
- docs/TOOLS.md — every operation per tool.
- docs/TROUBLESHOOTING.md — when things break.
- docs/reference/ — captured Inkscape 1.4.4
--action-listand--help.
Star history
Support
If you like my work and found this useful, feel free to support me at buymeacoffee.com/aravindev.
<a href="https://buymeacoffee.com/aravindev"> <img src="assets/buymeacoffee-qr.png" alt="Buy me a coffee QR code" width="180"> </a>
License
MIT — see LICENSE. This project is a fork of upstream MIT-licensed work; attribution and lineage details are in NOTICE.md.
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
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
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.