Introducing wrapture

Graham Dumpleton has put out Wrapture, a new Python library that takes the monkeypatching ideas behind wrapt and stretches them across testing and tracing at once. The pitch is simple enough: wrap a function or method, watch everything that passes through it, or swap in a different return value when needed.

That makes it useful in two directions. On one side, it works as an alternative to unittest.mock for tests. On the other, it gives you a way to attach tracing to code you don’t control, so you can record what flows through it without disturbing the program under observation.

Wrapture also includes OpenTelemetry support, plus a configuration-only setup for adding tracing to an existing Python project. Graham shows that with a compact config snippet that defines a summary capture, observes a Calculator target, and sends output to JSON lines in trace.jsonl.

The project is only a few weeks old, so this is early days. But there’s a more unusual detail here: every line of code and documentation in wrapture was written by an AI assistant under Graham’s direction. He is careful to draw a line between that and vibe coding, saying this was engineered deliberately by someone who already knew exactly what the result needed to be.