Generate shell completion

clipse ships a bash completion generator. Every CLI automatically gains a built-in generate-completion subcommand.

Generate and install it

$ mycli generate-completion

This does everything for you:

1. Writes the completion script to ~/.clipse.mycli.bash.
2. Adds source ~/.clipse.mycli.bash to your shell's startup file (once) so every new shell picks it up. The file is chosen from $SHELL: ~/.zshrc for zsh, otherwise ~/.bashrc.
3. Tells you the command to run to enable it in the current shell.

Completion script written to /home/you/.clipse.mycli.bash
Added "source /home/you/.clipse.mycli.bash" to /home/you/.zshrc
To enable completion in the current shell, run: source /home/you/.clipse.mycli.bash

The script is bash syntax, but it works under zsh too: it loads zsh's bashcompinit shim automatically when sourced from a zsh shell.

A running process cannot source into its parent shell, so the last step is left for you to run once in the current terminal:

source ~/.clipse.mycli.bash

Now pressing Tab completes your subcommands and options:

$ mycli <Tab>
sub   --help   --version   --opt   -o

Completion covers the first level (subcommands and the top-level options) and the second level (each subcommand's options).

Generate it programmatically

If you need the script as a string — to write it during an install step, for example — call generateCompletionScript():

import { writeFileSync } from "node:fs";
import { Clipse } from "clipse";

const cli = new Clipse("mycli", "…", "1.0.0").addSubcommands([/* … */]);
writeFileSync("mycli-completion.bash", cli.generateCompletionScript());

Unlike the generate-completion subcommand, generateCompletionScript() has no side effects: it neither prints nor exits.