A collection of structured skill files for Claude Code and other LLM tools. Each skill is a set of plain-text instructions and reference documents that load frameworks into an AI conversation, shaping how it thinks about a domain.
These aren't apps or libraries. They're config files — English-language instructions that turn a general-purpose AI into a more useful thinking partner for specific tasks.
A structured thinking partner for navigating medical information. Loads evidence-based medicine frameworks — PICO question framing, evidence hierarchy evaluation, statistics translation (relative vs. absolute risk, NNT), and appointment prep scaffolding — into a Claude conversation.
What it does: Helps you evaluate medical evidence and ask better questions. Not a diagnostician.
What it doesn't do: Diagnose, prescribe, or replace clinical judgment.
- Clone or download this repo
- Copy the skill directory into your Claude Code skills folder:
cp -r claude-code/medical-research ~/.claude/skills/medical-research - The skill activates automatically when you ask Claude Code about medical topics, or you can invoke it directly with
/medical-research
This repo uses dual licensing:
- Skill orchestration files (
SKILL.md) are licensed under the MIT License — use, modify, and redistribute freely. - Reference documents (
references/) are licensed under Creative Commons Attribution 4.0 International (CC BY 4.0) — use, adapt, and share freely with attribution.
See individual license files for full terms.
Built by Matthew @ idka.info — a blog called "I Don't Know Anything." These skills grew out of practical need: wanting an AI that helps me think more clearly about specific domains rather than just answering questions.
Blog post about the medical research skill: (coming soon)