AI Research Skills

AI Research Skills is an open, community-curated library of skills: small, teachable, composable units for doing research work with AI.

If you want the "why" and the bigger picture, read the Rationale.

What This Project Provides

A growing library of composable, teachable instructions for research workflows (from reference formatting to domain-specific methods).

Reproducible evaluations showing how different models perform on real research skills. Powered by BARS (Benchmark for AI Research Skills).

Practical ways to consume skills in real tools (e.g. Claude Code plugins, MCP workflows, CLI helpers), without requiring users to become Git experts.

Categories

Contributing

Want to add a skill? See our contribution guide for easy ways to contribute - no local Git setup required!

Workflow Stages

Skills are organized by where they fit in a typical research workflow:

StageDescription
discoverFinding sources, literature review
digitizeScanning, OCR, format conversion
extractText extraction, data scraping
annotateTagging, labeling, metadata enrichment
analyzeComputational analysis, statistics
modelMachine learning, topic modeling
connectLinking data, knowledge graphs
presentVisualization, publishing