Skip to content

Latest commit

 

History

History
77 lines (52 loc) · 3.24 KB

File metadata and controls

77 lines (52 loc) · 3.24 KB

Setup Guide for MedSci Skills

For physicians and clinical researchers who haven't used Python, R, or the command line before.

You don't need to be a programmer to use these skills, but you do need to install three things first: Claude Code, Python, and R. This guide walks you through it. If anything fails, run the diagnostic skill (/setup-medsci) inside Claude Code and it will tell you exactly what's missing.


Decision Tree

Your Computer Start Here
Mac (any model since 2017) mac.md
Windows 10/11 windows.md
Linux mac.md (the Homebrew section maps to apt/dnf — most commands transfer)

What You Will Install

Tool What It Does Required?
Claude Code The CLI that runs the skills Yes
Python 3.11+ Statistical analysis, figures, PDF processing Yes
R 4.x Meta-analysis (metafor), survival analysis, NHANES survey weighting Yes if you do MA / survival / survey
Node.js 20+ MCP servers (Zotero, Google Drive integration) Recommended
Git Cloning the repo, version control of your manuscripts Yes
Zotero + Better BibTeX Reference management; auto-sync to your manuscripts Yes for any paper

Time estimate: 30-60 minutes if everything goes smoothly. 1-2 hours if you hit an issue (most are documented in common-issues.md).


After You Install Everything

  1. Verify with the diagnostic skill (recommended):

    /setup-medsci
    

    This runs which python3, which Rscript, which claude, which node, and claude mcp list for you and prints a checklist showing what is ✅ and what is ❌.

  2. Install MedSci Skills (after the diagnostic shows green for the core tools):

    macOS — double-click installers/install-macos.command from the cloned repo.

    Windows — double-click installers/install-windows.cmd.

    Manually (any OS):

    git clone https://github.com/Aperivue/medsci-skills.git
    cd medsci-skills
    python3 installers/install.py --target claude
  3. Add MCP servers (optional but recommended for Zotero / Google Drive):

    See mcp-setup.md.

  4. Try Demo 1 to confirm everything works end-to-end:

    cd medsci-skills/demo/01_wisconsin_bc

    Then in Claude Code: /orchestrate --e2e (about 5-10 minutes; produces a manuscript, ROC curves, and a STARD compliance audit).


If Something Breaks

  • See common-issues.md — covers the top 10 issues physicians run into (PATH, Python 2 vs 3, Apple Silicon, antivirus, JSON syntax errors).
  • Run /setup-medsci again — the checklist will pinpoint the failure.
  • Open an issue at https://github.com/Aperivue/medsci-skills/issues with the /setup-medsci output and the error message.

Why a Setup Guide at All?

Most Claude Code skill collections assume you already have a working developer environment. We assume you don't — because most of our users are physicians and clinical researchers, and the install step is where they bounce. This guide aims to get you to a green diagnostic in under an hour, with no prior programming experience required.