If you've ever watched someone use an AI coding assistant, you might have noticed something a bit unnerving: it just... starts. No plan, no checks, no "wait, let me make sure I understand the problem first." It's fast, but it's also a bit like hiring a contractor who skips the measurements and goes straight to drilling holes in your wall.
Superpowers is an open-source project that changes that behaviour. You install it into tools like Claude Code or Cursor, and it adds what the authors call skills — mandatory routines the AI has to follow before and after any task. Think of it as a checklist that can't be skipped. Before writing a single line, the agent has to spec out what it's building. Before calling something fixed, it has to verify the fix actually works.
The community around it is growing fast — over 61,000 people have starred the project on GitHub, and new skills are being added every week.
If you have a developer (or a technical co-founder) using AI tools to build your product, this is the difference between an AI that produces a rough draft and one that produces something closer to finished work. Less back-and-forth, fewer surprises in production.
You don't need to install or touch anything yourself. But it's worth asking your dev: are you using something like Superpowers?
Coding agent — an AI assistant that doesn't just suggest code, but actually writes and runs it on your behalf, like a junior developer you can chat with.
Open-source — software that anyone can use, inspect, and improve for free. No licence fees, no vendor lock-in.
TDD (Test-Driven Development) — a habit where you write the test before the feature, so you always know when something is truly working. Superpowers can enforce this automatically.
Production-ready — code that's stable and safe enough to run on your live product, not just on a developer's laptop.