Open Data Products SDK v0.3.6

SDK how-to

Use this section as the practical website documentation for the SDK. It starts with basic CLI checks, then moves into generation, configuration, catalog and graph outputs, portfolio workflows, agent surfaces, recipes, OKF, Data Contracts, and Python usage.

Source of truth: commands on these pages are grounded in the local odps-python CLI and user docs. YAML remains the canonical standards artifact. HTML, TOON, and GCF outputs are review views or context sidecars.

  1. Create a clean Python virtual environment.
  2. Run open-data-products --version and validate one known file.
  3. Copy generation.config.yaml before editing provider settings.
  4. Generate one small artifact and validate it before building catalogs or portfolios.
  5. Use portfolio intake JSON when generated outputs look incomplete.
  6. Expose MCP only after the CLI works in the same environment.

Quick command map

Need Command Page
Validate a standards file open-data-products validate product.yaml Getting Started
Generate one signal or product reference open-data-products generate --input source_docs/signals/ --kind signal --output generated/ Generation
Build catalog and graph sidecars open-data-products odpc-build fragments/ --output catalog.yaml --toon catalog.toon --gcf catalog.gcf Reference Patterns
Build a portfolio workspace open-data-products portfolio build --objectives source_docs/objectives/ --products source_docs/products/ --output generated/portfolio/ Portfolio
Expose agent tools open-data-products serve Agents