A small Python library/framework for customising how the AST is walked and evaluated. I plan to use it in another project to configure a safe-ish sandboxed way of evaluating user submitted expressions and code. There are a few other libraries that enable this, but I found they bake in some stuff to the internals that I didn't want.