In linear algebra, eigen names what a transformation can’t turn

Apply a matrix to a vector and it usually ends up pointing somewhere new. An eigenvector is one of the exceptions. The matrix can make it longer or shorter, and it stays on the same line. The number that says how much longer is its eigenvalue, written with a lambda.

Coding agents are transformations too. They take a repository and return a different one, many times an hour. We wanted a tool that keeps the things that should survive that: which branch, whose request, what reason.

The word eigen is German for "own", which also fits an app that gives each thread its own folder, its own port and its own browser.

A long exposure: every star moves and the axis does not