CoReason-AI/coreason-manifest ? reverse-engineered prompt

Build me a Python package that acts as a strict source of truth for an ontology and manifest system.

I want it to be a pure data library, so importing it should not do anything active like starting processes, writing files, or making network calls. The models should be immutable, deterministic, and safe to hash the same way every time, even across different machines. Use structured validation for the schemas, and make sure the objects are easy to serialize to JSON in a canonical way.

Please include the core ontology models, the manifest schema, and tests that prove the package stays passive and produces stable output. Add clear docs for how someone would use it from Python, and keep the code organized so the main schema definitions are easy to find and maintain. If you need to check current docs for any library details, look them up online.