Agentic skills for the ModCon Base Data project. These skills extend coding agent tools (Claude Code, Codex, Goose etc.) with domain-specific workflows for ML dataset metadata and documentation.
| Skill | Description |
|---|---|
croissant-validator |
Validate and generate Croissant metadata (croissant.json) for ML datasets following the MLCommons Croissant 1.0 spec. |
datacard-generator |
Generate MODCON data cards for scientific datasets by introspecting a dataset directory. Supports three readiness levels: Discoverable (L1), Interoperable & Reusable (L2), and Understandable & Trustworthy (L3). |
hdmf-schema-builder |
Build HDMF (Hierarchical Data Modeling Framework) schema for organizing HDF5 data files for AI training and data sharing |
skills/
<skill-name>/
SKILL.md # Skill definition loaded by Claude Code
references/ # Supporting templates and reference documents (if any)
Skills are automatically available in Claude Code sessions when this repo is configured as a skills source. Invoke them by describing your task — Claude will select the appropriate skill — or reference them explicitly (e.g., "use the datacard-generator skill").