pydantic

Validate JSON payloads by defining a pydantic BaseModel with Field constraints, then calling model_validate_json on incoming data. Invalid inputs raise a ValidationError containing structured error details via errors().

Generate JSON Schema from pydantic BaseModel and TypeAdapter definitions to produce OpenAPI or LLM tool argument schemas. The output reflects your type hints, Field constraints, and discriminated unions for structured tool typing.

Discriminated unions validate polymorphic data by using a specific Field marker to route incoming JSON to the correct BaseModel subclass. This enables precise runtime validation and clear ValidationError reporting for untrusted payloads.

Parse environment variables using pydantic-settings, which maps environment configuration to Python type hints. It validates settings at runtime, applies strict coercion rules, and rejects malformed configuration inputs with actionable ValidationError details.

Strict mode rejects loosely-typed data to prevent unexpected coercion bugs during runtime validation. It enforces exact type matching on BaseModel fields, ensuring untrusted JSON payloads conform strictly to declared Python type hints.

Migrate from Pydantic v1 to v2 by replacing parse_obj and dict methods with model_validate and model_dump. Update field_validator and model_validator decorators to use explicit mode parameters for correct runtime validation and serialization.