Lambda abstractions can be defined using a backslash. \x.x, for example, denotes the identity function, where x is a variable. Note, that, since dropping the additional backslashes for multi-argument functions is allowed, \fx.fx is equivalent to \f.\x.fx.
Variables are usually single characters, but multi-character variables can be used by enclosing them in backquotes: `func`x
Lowercase and uppercase letters, as well as digits are allowed in variable names.
The interpreter allows for variables to be defined through let-bindings, such as let `id` = \x.x. These bindings all have to occupy their own line, but they don't necessarily have to be in order. The let-bindings can reference any other variables, as long as the definition doesn't end up being recursive.
These bindings are evaluated by detecting the free variables in a line and passing in the relevant bindings through a lambda. For example:
let `succ` = \nfx.f(nfx)
`succ`(\fx.x)
will result in
(\`succ`.`succ`(\fx.x))(\nfx.f(nfx))
being passed to the interpreter.