AI Skills for Group Sequential Design
Teaching AI to Design Clinical Trials
Overview
gsDesignSkills provides curated AI skills that give coding assistants domain expertise in clinical trial design. Each skill encodes package-specific knowledge, verified code patterns, and statistical conventions for R packages used in group sequential design, multiplicity control, and trial simulation.
Current version: 0.2.0
Skills are portable Markdown assets stored in .agents/skills/. Assistant- specific locations, such as .claude/skills/, are compatibility or export targets rather than the source of truth.
Quick Start
Use the portable skill source directly, or copy selected skills into the assistant-specific location your project expects:
# Copy all portable skills for a Claude Code project
cp -r gsDesignSkills/.agents/skills/* my-project/.claude/skills/
# Or copy a single skill
cp -r gsDesignSkills/.agents/skills/gsDesign my-project/.claude/skills/Then reference in your project’s CLAUDE.md:
## Skills
When working on group sequential designs,
read `.claude/skills/gsDesign/SKILL.md`.What’s Included
13 Skills Across 8+ R Packages
| Category | Skills |
|---|---|
| Group Sequential Design | gsDesign, gsDesign2, rpact |
| Multiplicity Control | graphicalMCP, gMCPLite, wpgsd |
| Trial Simulation | simtrial, gsDesignNB |
| Cross-Package | graphicalMCP-gsDesign2, illness-death, multi-endpoint-sim |
| Semantic Layer | semantic-router, trial-design-glossary |
Each Skill Contains
- SKILL.md — Trigger patterns, workflow, key concepts, and gotchas
- Code patterns — Tested, working code templates for common tasks
- llms.txt — Vendored machine-readable API documentation where available
Resources
- Presentation — Overview of the skills approach
- Simulation Example — Multi-endpoint trial simulation using 5 skills together
- Semantic Layer — Routing, glossary, and crosswalk assets
- GitHub Repository — Source code and contributions