Histopathology Annotation MCP
Annotate histopathology slides and images from chat, exporting to QuPath, COCO or YOLO
- Pricing
- Free
- Platforms
- CLI, Web
- Runs
- Locally
- Version
- 0.9.0
- Listing updated
- 3 days ago
About Histopathology Annotation MCP
Histopathology Annotation MCP is a free MCP server for design and media. It runs locally on your machine and is started by your AI client.
Built for pathology and computer-vision work, it connects an assistant to the Image Label annotation service. The assistant can upload an image, outline nuclei in an H&E slide with StarDist, trace regions from rough boxes with Segment Anything or MedSAM, check slide quality, normalise stain colours and clean up annotations. Output goes to COCO JSON, YOLO labels, YOLO classes or QuPath GeoJSON. The README says no GPU is needed, and it checks nuclei detection against expert outlines on the NuInsSeg dataset.
It works without an account: a guest identity is created and stored under ~/.config/label-image-mcp, images are kept for 7 days (up to 50) and there are 200 analyses a day. A free account, requested through the site, keeps images 90 days, lifts the daily cap and allows whole slides up to 5 GB. Each identity sees only its own images.
It installs with uvx from PyPI or as an .mcpb file for Claude Desktop, and a LangGraph agent is included.
Install
Run in a terminal
claude mcp add label-image-mcp --env LABEL_PASSWORD='<LABEL_PASSWORD>' -- uvx label-image-mcpAdd to claude_desktop_config.json
{
"mcpServers": {
"label-image-mcp": {
"command": "uvx",
"args": [
"label-image-mcp"
],
"env": {
"LABEL_PASSWORD": "<LABEL_PASSWORD>"
}
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"label-image-mcp": {
"command": "uvx",
"args": [
"label-image-mcp"
],
"env": {
"LABEL_PASSWORD": "<LABEL_PASSWORD>"
}
}
}
}Add to .vscode/mcp.json
{
"servers": {
"label-image-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"label-image-mcp"
],
"env": {
"LABEL_PASSWORD": "<LABEL_PASSWORD>"
}
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"label-image-mcp": {
"command": "uvx",
"args": [
"label-image-mcp"
],
"env": {
"LABEL_PASSWORD": "<LABEL_PASSWORD>"
}
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"label-image-mcp": {
"command": "uvx",
"args": [
"label-image-mcp"
],
"env": {
"LABEL_PASSWORD": "<LABEL_PASSWORD>"
}
}
}
}Configuration
LABEL_USERNAMEImage Label account username; leave unset to work as a guest (images kept 7 days, no whole slides)LABEL_PASSWORDImage Label account password (only with LABEL_USERNAME)LABEL_API_URLAPI base URL
Tools this server provides (1)
Taken from the project documentation; the running server may differ.
- detect_nuclei
- Outlines nuclei in H&E images; checked against NuInsSeg outlines.
What it does
- Nucleus detection
- Outlines every nucleus in an H&E slide automatically with StarDist.
- Assisted outlining
- Traces regions from rough boxes using Segment Anything or MedSAM.
- Slide quality checks
- Checks quality and normalises stain colours.
- Training-data export
- Exports COCO JSON, YOLO labels and classes, or QuPath GeoJSON.
- Guest mode
- Works without sign-up with images kept 7 days and 200 analyses a day.
- Whole-slide support
- Accounts can upload whole slides up to 5 GB.
Good for
- Turn an H&E slide into labelled nuclei for model training
- Export annotations to QuPath for review
- Check a slide's quality before analysis
- Trace tissue regions from rough boxes
- Prepare COCO or YOLO datasets from microscopy images
Pricing
Free
Free; guest mode limits apply, accounts free on request
Prices change. Check labels.sachin-kale.com before you buy.
Histopathology Annotation MCP FAQ
Do I need an account?
No. It runs as a guest, though guests cannot use whole slides.
How long are images kept?
7 days for guests (up to 50 images) and 90 days for accounts.
Does it need a GPU?
No, the README says no GPU is needed.
What export formats are supported?
COCO JSON, YOLO labels, YOLO classes and QuPath GeoJSON.
Is Histopathology Annotation MCP free?
Yes. Histopathology Annotation MCP is free to use.
How do I add Histopathology Annotation MCP to Claude Code?
Run this command in your terminal: claude mcp add label-image-mcp --env LABEL_PASSWORD='<LABEL_PASSWORD>' -- uvx label-image-mcp
What are the best alternatives to Histopathology Annotation MCP?
The closest alternatives to Histopathology Annotation MCP are HyperFrames by HeyGen, ffmpeg-skill, Claude Code Ultimate Guide, Tooluniverse and Oya Browser.
Alternatives to Histopathology Annotation MCP

HyperFrames by HeyGen
MCP serverHosted MCP connector for creating and rendering videos from HTML through AI chat
- Remote
- Freemium
- 85.3K GitHub stars
- Design and media

ffmpeg-skill
MCP serverLocal FFmpeg video and audio editing tools for coding agents over MCP
Runs locally- Open source
- 6.9K GitHub stars
- Design and media

Claude Code Ultimate Guide
MCP serverMCP server for searching the Claude Code Ultimate Guide from any MCP client
Runs locally- Open source
- 6.5K GitHub stars
- AI models and LLM tooling

Tooluniverse
MCP serverMCP server giving AI agents access to thousands of scientific research tools
Runs locally- Open source
- 1.8K GitHub stars
- Health and science

Oya Browser
MCP serverHIPAA-oriented agentic browser with replayable playbooks, reachable over MCP
- 32 tools
- Remote
- Freemium
- 349 GitHub stars
- Browser automation

Generate fictional character portraits and lip-synced talking videos from text over MCP
- 10 tools
- Remote
- Freemium
- Design and media