Slicer: a skill for code and 3D scenes
Find answers in 3D Slicer source and documentation, prepare a script and, with a separately connected MCP server, inspect a real scene using a public sample.
The complete kit in one ZIP
One ZIP: skills, instructions, installer and a public demo runner.
When to use it
Find APIs for a Slicer Python script or extension
Understand MRML scenes and view controls
Prepare a CTChest demonstration through the separate MCP server
What you’ll get
Code and documentation references and a verifiable action plan; with MCP available, actual scene information and a screenshot.
For example
Prepare Slicer using the pinned skill: first check the environment and choose web mode without indexes. Use only the public CTChest sample for a demonstration and report an unavailable MCP connection explicitly.
Before you start
Our editorial guide to Steve Pieper’s skill. MCP is a separate setup step; this release was source-reviewed.
Read the pinned SKILL.md and setup.sh. Keep the complete repository at that revision in a separate reference folder.
Register the skill using your agent’s supported mechanism. Start with web mode and no indexes; full mode involves substantial downloads.
For scene control, install 3D Slicer separately, review slicer-mcp-server.py and configure local MCP. It executes Python with Slicer privileges; use an isolated environment.
Check the connection and start with public CTChest data. Preserve actual scene information and screenshots; connection failures are not inspection results.
Original, requirements and limits
Original skill: Steve Pieper and contributors. Editorial guide: kossolapov.com.
- Agent
- An agent that can read Agent Skills instructions; MCP support is separately required for live Slicer control
- Tools
- Git, Bash and Perl for the setup helper; web mode avoids large source clones
- Installed 3D Slicer and its Python console for the separate slicer-mcp-server.py
- Optional Python search dependencies and disk space for BM25/hybrid indexes
- Possible costs
- The chosen AI host or remote compute may incur costs
- Limits
- Source-reviewed only: neither this skill nor its MCP server was run for this release.
- The MCP bridge can execute Python with the privileges of the Slicer process; use an isolated public-sample environment.
- MCP responses can reach a cloud AI host; a local Slicer window does not guarantee local-only data handling.
- Installing the skill does not install Slicer or connect MCP. The full setup can clone many repositories; choose its mode deliberately.
- Software control and screenshots are not clinical interpretation or a complete CT review.
- The pinned SKILL.md links to references/mcp.md, which is absent in that revision. Inspect the included slicer-mcp-server.py and verify MCP setup separately; the referenced setup guide is unavailable.