Plotter Studio
An MCP server that gives AI agents eyes and a robotic arm to compose and iterate generative art on an AxiDraw pen plotter via webcam feedback.
README
Plotter Studio
An MCP server that gives AI agents eyes and a robotic arm.
Plotter Studio connects AI agents to an AxiDraw pen plotter and a webcam via the Model Context Protocol, enabling the agent to compose generative art as SVG, send it to the plotter, observe the result through the camera, and iterate.
How it works
- The agent composes SVG artwork sized to your paper
- Plotter Studio sends the SVG to the AxiDraw via the NextDraw API
- The agent requests pen changes between passes (you do the swap)
- The agent captures a webcam frame to see the result
- The agent composes the next layer based on what it sees
- Repeat until the piece is done
Requirements
- uv (Python package manager)
- Python 3.13+
- AxiDraw or NextDraw pen plotter (NextDraw firmware)
- USB webcam or gphoto2-compatible camera (e.g. Sony A7III)
- Node.js 18+ (for mcp-remote)
- gphoto2 (optional, for tethered camera backend)
Quick start
1. Clone and install
git clone https://github.com/ucodia/plotter-studio-mcp.git
cd plotter-studio-mcp
uv sync
This creates a virtual environment and installs all dependencies (FastMCP, nextdraw-api, OpenCV, Pillow).
2. Set up a camera
Plotter Studio supports two camera backends: OpenCV (USB webcams) and gphoto2 (tethered cameras like Sony, Canon, Nikon).
Option A: USB webcam (default)
Plug in your webcam, then find its device index:
uv run python -c "
import cv2
for i in range(10):
cap = cv2.VideoCapture(i)
if cap.isOpened():
w, h = int(cap.get(3)), int(cap.get(4))
print(f' {i}: {w}x{h}')
cap.release()
"
Set CAMERA_INDEX to the index you want to use (default is 0).
Raspberry Pi note: The pip opencv-python-headless wheel may lack V4L2 support on ARM. If the camera is detected by lsusb but OpenCV can't open it, install the system package and recreate the venv with system site-packages:
sudo apt install python3-opencv
uv venv --system-site-packages --python python3
uv sync
Option B: gphoto2 (tethered camera)
For higher quality captures, you can use a tethered camera via gphoto2. This works with many Sony, Canon, and Nikon cameras connected over USB.
Install gphoto2:
# macOS
brew install gphoto2
# Debian/Ubuntu/Raspberry Pi OS
sudo apt install gphoto2
Set your camera to PC Remote USB mode (the exact menu path varies by model), connect it via USB, and verify detection:
gphoto2 --auto-detect
Then set CAMERA_BACKEND=gphoto2 in your .env file. Captures will be slower (a few seconds per shot) but image quality will be significantly better. Set your camera to JPG-only (no RAW) to speed up transfers.
3. Run the server
uv run plotter-studio
This starts the MCP server over SSE at http://127.0.0.1:8888/sse. To listen on all interfaces (e.g. when running on a remote Pi), pass --host:
uv run plotter-studio --host
4. Configure Claude Desktop
Open your Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add the plotter-studio server:
{
"mcpServers": {
"Plotter Studio": {
"command": "npx",
"args": [
"mcp-remote@latest",
"http://127.0.0.1:8888/sse",
"--allow-http"
]
}
}
}
5. Restart Claude Desktop
Quit and reopen Claude Desktop. You should see Plotter Studio listed as a connected MCP server (look for the hammer icon in the chat input area). If it doesn't appear, check the MCP logs in Claude Desktop's developer console.
6. Test it
Ask Claude something like:
What's the plotter status?
or
Capture a photo from the camera so I can see the paper.
If both respond without errors, you're ready to make art.
Tools
| Tool | What it does |
|---|---|
server_info |
Get HTTP base URL and file transfer endpoints |
plot_start |
Plot an uploaded SVG by file ID (background, non-blocking) |
plot_stop |
Cancel the current plot gracefully |
plot_status |
Check plotter state (idle/plotting/error) |
capture |
Take a photo, returns file reference for HTTP download |
tool_move |
Move tool to a position (tool up) |
tool_raise |
Raise the tool |
tool_home |
Return tool carriage to home (0,0) |
notify |
Send a notification to the human operator |
Environment variables
All configuration is via environment variables:
| Variable | Default | Description |
|---|---|---|
SVG_DIR |
output |
Directory where SVG files are saved |
WEBHOOK_URL |
(empty) | Webhook URL for push notifications (supports ntfy.sh) |
PLOTTER_MODEL |
2 |
NextDraw model number (2 = AxiDraw V3/A3) |
PLOTTER_PENLIFT |
3 |
Pen lift type (3 = brushless servo) |
PLOTTER_PEN_POS_DOWN |
0 |
Pen-down servo position as percentage (0=lowest) |
PLOTTER_PEN_POS_UP |
50 |
Pen-up servo position as percentage (100=highest) |
CAMERA_BACKEND |
opencv |
Capture backend: opencv for webcams, gphoto2 for tethered cameras |
CAMERA_INDEX |
0 |
Webcam device index (opencv backend only) |
JPEG_QUALITY |
90 |
JPEG output quality (1-100) |
CAMERA_ROTATE_LANDSCAPE |
0 |
Camera rotation in degrees for landscape orientation |
CAMERA_ROTATE_PORTRAIT |
90 |
Camera rotation in degrees for portrait orientation |
MCP_PORT |
8888 |
Port for the MCP SSE server |
HTTP_BASE_URL |
http://localhost:{MCP_PORT} |
Base URL for HTTP file transfer endpoints |
Notifications
Plotter Studio can send push notifications so you don't have to watch the screen while it plots. Set WEBHOOK_URL to a ntfy.sh topic URL and subscribe on your phone. You'll get pinged when a plot starts, finishes, errors, or when the agent needs you to swap a pen.
Generic JSON webhooks also work. If the URL doesn't contain "ntfy", the server POSTs a JSON body with event, timestamp, and event-specific fields.
Project structure
src/plotter_studio/
server.py # MCP server, tool definitions, HTTP routes, config
plotter.py # AxiDraw control and state machine
camera.py # Webcam and gphoto2 capture
filestore.py # Temp file store for HTTP file transfers
webhook.py # Push notifications
tests/
test_plotter_state.py
License
AGPL-3.0-or-later
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.