WPL Ecosystem

The canonical home for the WPL specification, JSON Schema (Draft 2020-12), reference examples, and the conformance test suite. This repo is the source of truth: every implementation is verified against the conformance fixtures here.

Reference TypeScript validator. Two-pass design: JSON Schema validation (Draft 2020-12 via Ajv) followed by semantic invariants (duplicate-ID detection, ref resolution against your catalog, prescription validity, personalization rule shape, etc.). Returns structured ValidationError[] with RFC 6901 JSON Pointer paths so you can render findings in any UI.

The compiler that turns the human/LLM-friendly WPL-AI DSL into canonical WPL JSON. Includes lexer, parser, compiler, DSL-level semantic warnings, and a JSON-pointer-to-source-range map for editor integrations. The same compiler that runs in the Playground.

How the pieces fit together

Quick path: author plans in WPL-AI (or have an LLM emit it), compile to JSON, validate against the schema. Three packages, one canonical format.

Naming & trademark

"WPL" and "Wellness Plan Language" are trademarks of Gymbile. Implementations that pass the conformance suite may declare themselves "WPL-compatible" or "WPL v1 compliant" — encouraging interop while keeping the name itself stable. See the trademark policy for the full text.

Contributing

We're actively maintaining all three repos. Issues, PRs, and discussions are welcome:

• Spec changes — open an issue first on gymbile/wpl so we can discuss before you spend time on a PR.
• New error rules — propose them on the schema repo with a conformance fixture; both validators implement.
• Compiler / DSL improvements — file on gymbile/wpl-ai.
• Validator implementations in other languages — talk to us; we want to coordinate naming and conformance.