nnU-Net (Claude Skill)

Medical image segmentation with nnU-Net’s self-configuring framework — auto-selects architecture, preprocessing, training for any modality.


Type	Claude Skill
Supplier	jaechang-hits (community OSS, CC BY 4.0)
Availability	GA — part of the BixBench-evaluated SciAgent-Skills collection
Pricing	Free / OSS (Apache-2.0)
Capabilities	Read/Write — Claude runs the skill’s Python locally (Bash), not as an MCP tool

How to install

SciAgent-Skills is not an npm package — skills are plain markdown read directly by the agent (no npx/npm).

Claude Code — clone and load as a plugin:
```
git clone https://github.com/jaechang-hits/SciAgent-Skills
```
Then inside Claude Code run /plugin install sciagent-skills (verify it appears under /plugin → Installed). Clone into your project directory so Claude Code picks the skills up via CLAUDE.md.
Manual / other agents — point the agent at the skill file directly:
```
cp -r SciAgent-Skills/skills/medical-imaging/nnunet-segmentation ~/.claude/skills/
```
The skill declares its own Python dependencies in its SKILL.md; install them when prompted on first use.

What it does

Medical image segmentation with nnU-Net’s self-configuring framework — auto-selects architecture, preprocessing, training for any modality. CT, MRI, microscopy, ultrasound in 2D, 3D full-res, 3D low-res, cascade. Pipeline: convert → plan/preprocess → train (5-fold CV) → best config → predict → ensemble. Use when classical segmentation fails and annotated data exists.

Primary use cases: classical segmentation fails and annotated data exists.

Notes

Distributed as a SKILL.md (plus code examples) in the SciAgent-Skills collection — Claude executes it locally via Bash/Python rather than as an MCP server. Upstream license: Apache-2.0. The skill directory upstream is skills/medical-imaging/nnunet-segmentation.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.