cofin/dishka

Dishka Dependency Injection Skill

Overview

Dishka is a Python dependency injection framework built around Providers, Scopes, and typed containers. It supports async/sync workflows and integrates with web frameworks (Litestar, FastAPI) and CLI tools (Click).

---

<workflow>

References Index

For detailed guides and configuration examples, refer to the following documents in references/:

• Providers, Scopes & Factory Functions
  • Core concepts, scope hierarchy, container creation, provider patterns, clean naming, and best practices.
• Litestar Integration
  • Setup, controller injection, router integration, and manual resolution from connection.
• FastAPI Integration
  • Setup and route-level injection with FromDishka.
• CLI Integration
  • Click with async_inject decorator for Dishka-powered CLI commands.
• Testing Patterns
  • Test containers, mock providers, and override strategies.

</workflow> <example>

Example: Provider and Container Setup

from dishka import Provider, Scope, make_async_container, provide

class AppProvider(Provider):
    scope = Scope.APP

    @provide
    async def get_db_engine(self) -> AsyncEngine:
        return create_async_engine("postgresql+asyncpg://...")

class RequestProvider(Provider):
    scope = Scope.REQUEST

    @provide
    async def get_session(self, engine: AsyncEngine) -> AsyncSession:
        return AsyncSession(engine)

container = make_async_container(AppProvider(), RequestProvider())

</example>

---

Official References

• https://dishka.readthedocs.io/en/stable/
• https://dishka.readthedocs.io/en/stable/integrations/litestar.html
• https://dishka.readthedocs.io/en/stable/integrations/fastapi.html
• https://dishka.readthedocs.io/en/stable/integrations/click.html
• https://github.com/reagento/dishka/releases
• https://pypi.org/project/dishka/

Shared Styleguide Baseline

• Use shared styleguides for generic language/framework rules to reduce duplication in this skill.
• General Principles
• Dishka
• Python
• Keep this skill focused on tool-specific workflows, edge cases, and integration details.

<guardrails> ## Guardrails

• Explicitly manage Scopes (APP, REQUEST) -- Always use the appropriate scope to avoid resource leaks or unnecessary object creation. Objects in Scope.APP live as long as the container; Scope.REQUEST lives only for the duration of a request.
• Avoid global container access -- Always use dependency injection to provide dependencies; never resolve objects from a global container instance in application logic.
• Ensure Providers are stateless -- Providers should only contain factory methods; any state should be managed within the injected objects themselves.
• Check scope hierarchy -- Objects in a wider scope (APP) cannot depend on objects in a narrower scope (REQUEST).
• Use typed providers -- Always use type hints for provider return values to ensure the container can correctly resolve and validate dependencies. </guardrails>

<validation> ## Validation Checkpoint

• [ ] Providers are assigned the correct Scope (APP, REQUEST)
• [ ] No objects are resolved manually from a global container
• [ ] All factory methods in providers are correctly annotated with @provide
• [ ] Scope hierarchy is valid (no narrow-to-wide scope dependencies)
• [ ] Provider return types match the types expected by the consumers
• [ ] Async/sync providers are used consistently with the target framework </validation>