Shell integration

Why it exists

Some generated commands change your shell’s state rather than doing work — cd into a directory, export a variable. A plain subprocess cannot do that for you: it changes its own environment and then exits.

smartly ships a shell function that runs the generation in a subprocess, then evaluates the result in your shell. Add one line to your shell startup file:

# ~/.zshrc
eval "$(smartly init zsh)"

# ~/.bashrc
eval "$(smartly init bash)"

Both shells are supported on both platforms — pick the one you actually run:

eval "$(smartly init bash)"   # bash
eval "$(smartly init zsh)"    # zsh

Open a new shell, or source the file, and you are done.

What changes, and what does not

Without the shell function, smartly still works for everything that does not mutate your shell’s state, which is most commands. A generated cd or export will run — it just affects the subprocess it ran in, not your interactive shell.

How it works

The function calls smartly --print-only, which generates and prints the command without running it, and then evals that output in the parent shell. Once the command finishes, the function calls smartly --record-exit with the exit code so the outcome lands in your history log.

--print-only and --record-exit are internal. They are documented here because you will see them if you read the emitted function, but they are not meant to be typed by hand.

Inspect what gets sourced at any time:

smartly init zsh

It prints the function to stdout — that is what eval consumes.