Open source

Skills we use, published as MIT

Every entry is a note we actually ran into, packaged so an agent can read it and you can copy it. One repo each, MIT, no signup.

3 repos

What these repositories are

Each one started as a real failure: we hit it in our own system, worked out the mechanism, then extracted the smallest reproducible piece and made it a standalone repository. They are not appendix material and not a reference implementation - each runs on its own, and what it prints is what a page on this site quotes.

Why MIT

MIT is the one license that does not require reading a legal document: use it, change it, put it in commercial work, keep the notice. We avoid copyleft licences because these repos exist to be used inside other people's projects, and we avoid vague phrases like for learning only, which grant nothing at all.

How we mark verification

Every entry carries reproduced / partly reproduced / not reproduced plus the date we checked, and that is our own label rather than a star count. Each detail page states the problem, the stack, how to install it, and which page on this site used it. If an entry says not reproduced, it ran in exactly one environment - do not read it as a general guarantee.

What never goes in here

Business logic from our internal systems, rate and deduction models, customer data structures. What gets open-sourced is the mechanism, not the production detail.