Technical Blog

Like most Python developers, I spin up new environments for different projects: different dependencies, different Python versions, little experiments I don't want to contaminate. At the time, it felt organized.

I assumed I had maybe 5 or 6 environments active.

When I finally checked, I had 26 scattered across Conda, venv, uv and Mamba, together chewing up ~45GB on my Windows machine. On my Mac, which I thought was "clean," I found another 4 using ~5GB.

Some of those weren't even mine directly — a couple came from AI coding agents I was testing, which silently created new virtual environments in the background...

Last month I discovered 26 dormant Python environments silently eating 45GB on my machine. That experience led me to build PyEnvManager, which I introduced in my first post.

At first, I thought this was just my bad housekeeping. But the 2025 State of Python survey by JetBrains and the PSF makes it clear: I'm not alone. Environment sprawl and dependency clutter are among the most common pain points across the Python ecosystem.

According to the survey data: 51% of Python developers use it for Data Science & AI, 83% are not on the latest Python version, and 68% value tools that reduce manual environment management...