Kyzo

Kyzo's skills guide construction of disciplined Python model code: five proven domain data shapes, verb and derivation methods, consistency models and bindings for live state, config/route/composition-root plumbing, and foreign-model/contract-model/ordered-union boundary crossings. Companion failure skills flag anti-patterns like stub bodies, bare primitives, and hand-rolled exception handling.

Python engineers building Pydantic-style domain models who want enforced architectural boundaries. The skills target developers writing model methods, mutable state holders, transport clients, route handlers, settings readers, and foreign payload adapters, ensuring every construct has a modeled declaration authorizing it.

Each success skill defines the legal constructs for a layer — values, verbs, state, wiring, adapters — while its paired failure skill fires when code is about to violate that layer, such as an undecorated method on a frozen model, a top-level def outside route files, or a union built without a discriminator.

The skills assume a Python codebase using Pydantic-style models with frozen models, validators, discriminated unions, and typed fields. Concepts referenced include consistency models, composition roots, route files, transport clients, and environment configuration reads, so familiarity with layered application architecture is expected.

Kyzo is a pure-Rust unified retrieval database combining relational, graph, vector, text, and time-series data in a single Datalog engine. The published skills complement this by enforcing rigorous domain modeling and boundary-crossing discipline in Python programs that interact with such data systems.