Documentation

Typed dependency injection for small Python apps.

Injex wires services, CLIs, workers, and tests with regular Python type hints, explicit registrations, scoped lifetimes, overrides, and startup validation.

Python 3.10–3.14 Zero runtime dependencies PEP 561 typed

Start Getting started Install, register services, validate the graph, resolve use cases. Safety Graph validation Find missing registrations and cycles before services are built. Decision Comparison guide Choose between manual wiring, framework DI, Injex, and larger containers. Speed Performance notes Benchmark shape, local results, and reproduction command. Boundary Compared to FastAPI Depends Use FastAPI at the HTTP edge and keep app wiring reusable. Async Async resolution Async factories, async-resource lifecycles, and per-request async scopes. Trade-offs Compared to DI frameworks Manual wiring, Injex, Wireup, Dishka, and dependency-injector. Recipes Application wiring recipes FastAPI composition root, worker job scope, CLI commands, and test overrides. Migration Migrating from factories Move a hand-written factories.py to Injex without changing application classes. Reference API reference Container methods, scopes, overrides, validation, and exceptions. Guide Python dependency injection Understand when DI helps in Python and how Injex stays small. Article Why tiny Python DI An article about explicit wiring for Python services. Hashnode Where should dependency wiring live? FastAPI, Typer, workers, tests, and the app composition root. DEV Fast dependency injection in Python Injex 1.3.0, benchmark context, and framework boundaries.

Install

pip install injex

Best fit

Service layers reused by API handlers, CLIs, workers, and tests.
Clean architecture code that should not depend on a web framework.
Tests that need temporary dependency overrides.
Startup checks that should validate the dependency graph early.

Rule of thumb: keep container calls at the composition root. Application services should depend on normal Python types.