Claude Skills for Scientific Notebooks

Seven skills for AI-assisted oceanographic data development.

The Setup

The C4IROcean-OceanDataPlatform repo has seven Claude skills in .claude/skills/. Each teaches Claude a specific domain.

The Skills

1. stac-api-reference

How to search and access STAC catalogs. Includes common queries, authentication patterns, and asset loading.

2. odp-dataset-patterns

Ocean Data Platform specifics: available collections, data formats, coordinate systems.

3. hub-ocean-odp

Navigation for the ODP Hub: finding datasets, understanding metadata, data lineage.

4. dask-odp-patterns

Chunking strategies, lazy loading, distributed processing for large ocean datasets.

5. notebook-formatting

Cell structure, markdown conventions, visualization standards for reproducible notebooks.

6. odp-troubleshooting

Common errors and fixes: authentication, timeouts, memory issues, coordinate mismatches.

7. proposal-writing

Grant and project proposal templates, boilerplate for ocean data projects.

Why It Works

Each skill is:

• Focused — One topic per file
• Searchable — Named after keywords Claude will encounter
• Practical — Code snippets, not theory

When you say “load SST data from ODP,” Claude finds hub-ocean-odp and odp-dataset-patterns. When you hit a memory error, dask-odp-patterns and odp-troubleshooting activate.

Building Your Own

1. Identify repeated explanations in your notebooks
2. Extract to .claude/skills/topic-name.md
3. Include code snippets that work
4. Let Claude write the skill from your examples

---

Domain expertise, always available.

See the skills