pardesline
Python 3D vision and medical imaging datasets and algorithms, served to agents over MCP
- Pricing
- Open source
- Platforms
- Linux, macOS, Windows
- Runs
- Locally
- Version
- 0.5.0
- Listing updated
- 3 days ago
About pardesline
pardesline is an open-source MCP server for health and science. It runs locally on your machine and is started by your AI client.
pardesline is a Python package that loads public 3D and medical datasets and runs algorithms on point clouds, meshes and CT or MRI volumes. Its MCP server runs on your machine without an account and exposes the same catalogue: 43 datasets and 73 algorithms, through fourteen tools, five resources and a start_here prompt.
When the package cannot know something, such as a licence, units or array layout, it returns the question as JSON for the agent to pass to you instead of guessing. Datasets are described in manifests that carry files, labels, frame, units, licence and citation. Twenty bundled skills, including registration, PointNet segmentation, Gaussian splatting and atlas label transfer, are served through list_skills and get_skill.
Install with pip install "pardesline[mcp]" and start it with pardesline mcp over stdio or streamable HTTP. A Docker image and a uvx recipe are also documented.
Install
Run in a terminal
claude mcp add pardesline --env HF_TOKEN='<HF_TOKEN>' -- uvx pardeslineAdd to claude_desktop_config.json
{
"mcpServers": {
"pardesline": {
"command": "uvx",
"args": [
"pardesline"
],
"env": {
"HF_TOKEN": "<HF_TOKEN>"
}
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"pardesline": {
"command": "uvx",
"args": [
"pardesline"
],
"env": {
"HF_TOKEN": "<HF_TOKEN>"
}
}
}
}Add to .vscode/mcp.json
{
"servers": {
"pardesline": {
"type": "stdio",
"command": "uvx",
"args": [
"pardesline"
],
"env": {
"HF_TOKEN": "<HF_TOKEN>"
}
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"pardesline": {
"command": "uvx",
"args": [
"pardesline"
],
"env": {
"HF_TOKEN": "<HF_TOKEN>"
}
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"pardesline": {
"command": "uvx",
"args": [
"pardesline"
],
"env": {
"HF_TOKEN": "<HF_TOKEN>"
}
}
}
}Configuration
PARDESLINE_CACHEDownload cache (default ~/.cache/pardesline).HF_TOKENHugging Face token, only for gated datasets whose terms you accepted (AbdomenAtlas, CT-RATE).
Tools this server provides (14)
Taken from the project documentation; the running server may differ.
- list_datasets
- List available datasets.
- describe_dataset
- Describe one dataset.
- list_samples
- List samples in a dataset.
- load_sample
- Load a sample.
- scan_dataset
- Scan a private folder into a dataset.
- confirm_dataset
- Confirm a scanned dataset.
- export_dataset
- Export a dataset.
- list_methods
- List algorithms.
- describe_method
- Describe an algorithm.
- run_method
- Run an algorithm.
- describe_object
- Describe a loaded object.
- to_mesh
- Convert an object to a mesh.
- list_skills
- List bundled expert skills.
- get_skill
- Fetch one skill.
What it does
- 43 datasets
- Public 3D and medical datasets described with licences and citations.
- 73 algorithms
- Registration, segmentation and other methods with defaults tied to the data.
- Questions returned as JSON
- Licence, units and layout questions go back to the human instead of being guessed.
- Expert skills
- Twenty skills with reference scripts are served to any MCP client.
- Private folder scanning
- Scan your own data with scan_dataset and confirm_dataset.
- Several launch options
- Run over stdio, streamable HTTP, uvx or Docker.
Good for
- Register a CT and MR scan pair
- Browse a medical dataset and read its licence
- Run a point-cloud registration method on a sample
- Convert a volume to a mesh
- Get an agent to follow an expert registration or segmentation guide
Pricing
Open source
Prices change. Check pardesline.github.io before you buy.
pardesline FAQ
Does it need an account?
No. The MCP server runs on your machine without an account.
Which clients are documented?
Claude Code, Claude Desktop, Cursor, Gemini CLI and Codex CLI.
Are gated datasets supported?
Yes, with a Hugging Face token for gated sets whose terms you accepted, such as AbdomenAtlas and CT-RATE.
How many tools does it expose?
Fourteen tools, five resources and a start_here prompt.
Is pardesline free?
Yes. pardesline is open source, so you can use and self-host it for free.
How do I add pardesline to Claude Code?
Run this command in your terminal: claude mcp add pardesline --env HF_TOKEN='<HF_TOKEN>' -- uvx pardesline
What are the best alternatives to pardesline?
The closest alternatives to pardesline are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.
Alternatives to pardesline

REA
MCP serverCLI and MCP server that lets coding agents reverse engineer apps and binaries
Runs locally- Open source
- 464K GitHub stars
- Developer tools

Codebase Memory
MCP serverLocal MCP server that builds a code knowledge graph for AI coding agents
Runs locally- Open source
- 48.7K GitHub stars
- Knowledge and memory

GitHub
MCP serverOfficial MCP server connecting AI assistants to GitHub repos, issues and workflows
Not checked yet- Remote
- Open source
- 34.5K GitHub stars
- Developer tools

Serena MCP: the IDE for your agent
MCP serverMCP toolkit giving coding agents symbol-level code retrieval and editing
Runs locally- Open source
- 31.5K GitHub stars
- Developer tools

XERJ
MCP serverLocal search engine for AI agents with cited answers over your code and docs
Runs locally- Open source
- 19.2K GitHub stars
- Knowledge and memory

Ouroboros
MCP serverSpec-first layer that pins requirements and verifies results of AI coding agent work
Runs locally- Open source
- 6.4K GitHub stars
- Developer tools