docs: add util.parseArgs replacements for CLI argument parsers

Adds migration documentation for replacing minimist, mri, arg, meow,
yargs-parser, yargs, commander, and sade with Node.js built-in
util.parseArgs (available in Node 18.3+/16.17+).

Author: Benjamin Newman

docs/modules/parseargs.md

@@ -0,0 +1,387 @@
+---
+description: Modern alternatives to CLI argument parsing packages using Node.js built-in util.parseArgs
+---
+
+# Replacements for argument parsers
+
+Node.js 18.3+ (and 16.17+) includes [`util.parseArgs`](https://nodejs.org/api/util.html#utilparseargsconfig), a built-in argument parser that can replace many common CLI parsing libraries.
+
+## Packages covered
+
+- `minimist`
+- `mri`
+- `arg`
+- `meow`
+- `yargs-parser`
+- `yargs`
+- `commander`
+- `sade`
+
+## `minimist`
+
+`minimist` is the simplest migration case. It's often a transitive dependency rather than a direct choice:
+
+```ts
+import minimist from 'minimist' // [!code --]
+import { parseArgs } from 'node:util' // [!code ++]
+
+const argv = minimist(process.argv.slice(2)) // [!code --]
+const { values, positionals } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    force: { type: 'boolean', short: 'f' }, // [!code ++]
+    output: { type: 'string', short: 'o' }, // [!code ++]
+  }, // [!code ++]
+  allowPositionals: true, // [!code ++]
+}) // [!code ++]
+
+// Access options
+argv.force // [!code --]
+values.force // [!code ++]
+
+// Access positionals
+argv._ // [!code --]
+positionals // [!code ++]
+```
+
+### Handling unknown options
+
+`minimist` accepts any flag by default. To match this behavior, use `strict: false`:
+
+```ts
+const { values, positionals } = parseArgs({
+  args: process.argv.slice(2),
+  strict: false,
+  allowPositionals: true,
+})
+```
+
+### Providing a minimist-compatible interface
+
+For gradual migration, you can create a compatibility layer:
+
+```ts
+import { parseArgs } from 'node:util'
+
+const { values, positionals } = parseArgs({
+  args: process.argv.slice(2),
+  options: {
+    help: { type: 'boolean', short: 'h' },
+    force: { type: 'boolean', short: 'f' },
+  },
+  strict: false,
+  allowPositionals: true,
+})
+
+// minimist-compatible object
+const argv = {
+  _: positionals,
+  ...values,
+  // Add short aliases only when truthy (minimist behavior)
+  ...(values.help && { h: values.help }),
+  ...(values.force && { f: values.force }),
+}
+```
+
+### Using tokens for advanced parsing
+
+For minimist-style parsing where `--flag value` treats `value` as the flag's argument (not a positional), use the `tokens` option:
+
+```ts
+const { tokens } = parseArgs({
+  args: process.argv.slice(2),
+  strict: false,
+  allowPositionals: true,
+  tokens: true,
+})
+
+const result = { _: [] }
+for (let i = 0; i < tokens.length; i++) {
+  const token = tokens[i]
+  if (token.kind === 'option') {
+    const nextToken = tokens[i + 1]
+    // Check if boolean flag is followed by a value
+    if (token.value === undefined && nextToken?.kind === 'positional') {
+      result[token.name] = nextToken.value
+      i++ // Skip next token
+    } else {
+      result[token.name] = token.value ?? true
+    }
+  } else if (token.kind === 'positional') {
+    result._.push(token.value)
+  }
+}
+```
+
+## `mri`
+
+`mri` is a lightweight minimist alternative. The migration is nearly identical to minimist:
+
+```ts
+import mri from 'mri' // [!code --]
+import { parseArgs } from 'node:util' // [!code ++]
+
+const argv = mri(process.argv.slice(2), { // [!code --]
+  alias: { h: 'help', v: 'version' }, // [!code --]
+  boolean: ['help', 'version'], // [!code --]
+}) // [!code --]
+const { values, positionals } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    help: { type: 'boolean', short: 'h' }, // [!code ++]
+    version: { type: 'boolean', short: 'v' }, // [!code ++]
+  }, // [!code ++]
+  allowPositionals: true, // [!code ++]
+}) // [!code ++]
+```
+
+## `arg`
+
+`arg` uses a schema-based approach similar to `parseArgs`:
+
+```ts
+import arg from 'arg' // [!code --]
+import { parseArgs } from 'node:util' // [!code ++]
+
+const args = arg({ // [!code --]
+  '--port': Number, // [!code --]
+  '--host': String, // [!code --]
+  '--verbose': Boolean, // [!code --]
+  '-p': '--port', // [!code --]
+  '-h': '--host', // [!code --]
+}) // [!code --]
+const { values } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    port: { type: 'string', short: 'p' }, // [!code ++]
+    host: { type: 'string', short: 'h' }, // [!code ++]
+    verbose: { type: 'boolean' }, // [!code ++]
+  }, // [!code ++]
+}) // [!code ++]
+
+// Note: parseArgs returns strings, convert if needed
+const port = Number(values.port) // [!code ++]
+```
+
+> [!NOTE]
+> `parseArgs` only supports `string` and `boolean` types. For numbers, parse the string value yourself.
+
+## `meow`
+
+`meow` is popular for small CLIs, combining parsing with auto-help from package.json:
+
+```ts
+import meow from 'meow' // [!code --]
+import { parseArgs } from 'node:util' // [!code ++]
+import { readFileSync } from 'node:fs' // [!code ++]
+
+const cli = meow(` // [!code --]
+  Usage // [!code --]
+    $ my-cli <input> // [!code --]
+
+  Options // [!code --]
+    --rainbow, -r  Include a rainbow // [!code --]
+    --postfix      Append a string // [!code --]
+`, { // [!code --]
+  importMeta: import.meta, // [!code --]
+  flags: { // [!code --]
+    rainbow: { type: 'boolean', shortFlag: 'r' }, // [!code --]
+    postfix: { type: 'string', default: '!' }, // [!code --]
+  } // [!code --]
+}) // [!code --]
+cli.input // => positionals // [!code --]
+cli.flags // => { rainbow: false, postfix: '!' } // [!code --]
+
+const { values, positionals } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    rainbow: { type: 'boolean', short: 'r' }, // [!code ++]
+    postfix: { type: 'string' }, // [!code ++]
+    help: { type: 'boolean', short: 'h' }, // [!code ++]
+    version: { type: 'boolean', short: 'v' }, // [!code ++]
+  }, // [!code ++]
+  allowPositionals: true, // [!code ++]
+}) // [!code ++]
+const postfix = values.postfix ?? '!' // [!code ++]
+
+// Handle --help yourself // [!code ++]
+if (values.help) { // [!code ++]
+  console.log(` // [!code ++]
+  Usage // [!code ++]
+    $ my-cli <input> // [!code ++]
+
+  Options // [!code ++]
+    --rainbow, -r  Include a rainbow // [!code ++]
+    --postfix      Append a string // [!code ++]
+`) // [!code ++]
+  process.exit(0) // [!code ++]
+} // [!code ++]
+
+// Handle --version yourself // [!code ++]
+if (values.version) { // [!code ++]
+  const pkg = JSON.parse(readFileSync(new URL('./package.json', import.meta.url), 'utf8')) // [!code ++]
+  console.log(pkg.version) // [!code ++]
+  process.exit(0) // [!code ++]
+} // [!code ++]
+```
+
+> [!NOTE]
+> `meow` provides automatic `--help` and `--version` handling from your package.json. With `parseArgs`, you implement these yourself. For very simple CLIs, this trade-off may not be worth it.
+
+## `yargs-parser`
+
+`yargs-parser` (the parsing engine behind `yargs`) has more features, but basic usage maps directly:
+
+```ts
+import yargsParser from 'yargs-parser' // [!code --]
+import { parseArgs } from 'node:util' // [!code ++]
+
+const argv = yargsParser(process.argv.slice(2), { // [!code --]
+  alias: { h: 'help' }, // [!code --]
+  boolean: ['help', 'verbose'], // [!code --]
+  string: ['config'], // [!code --]
+}) // [!code --]
+const { values, positionals } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    help: { type: 'boolean', short: 'h' }, // [!code ++]
+    verbose: { type: 'boolean' }, // [!code ++]
+    config: { type: 'string' }, // [!code ++]
+  }, // [!code ++]
+  allowPositionals: true, // [!code ++]
+}) // [!code ++]
+```
+
+## `yargs`
+
+`yargs` uses a chained builder API. For simple cases without subcommands:
+
+```ts
+import yargs from 'yargs' // [!code --]
+import { hideBin } from 'yargs/helpers' // [!code --]
+import { parseArgs } from 'node:util' // [!code ++]
+
+const argv = yargs(hideBin(process.argv)) // [!code --]
+  .option('port', { // [!code --]
+    alias: 'p', // [!code --]
+    type: 'number', // [!code --]
+    default: 3000, // [!code --]
+  }) // [!code --]
+  .option('host', { // [!code --]
+    alias: 'h', // [!code --]
+    type: 'string', // [!code --]
+    default: 'localhost', // [!code --]
+  }) // [!code --]
+  .option('verbose', { // [!code --]
+    type: 'boolean', // [!code --]
+    default: false, // [!code --]
+  }) // [!code --]
+  .parseSync() // [!code --]
+
+const { values } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    port: { type: 'string', short: 'p' }, // [!code ++]
+    host: { type: 'string', short: 'h' }, // [!code ++]
+    verbose: { type: 'boolean' }, // [!code ++]
+  }, // [!code ++]
+}) // [!code ++]
+const port = Number(values.port ?? '3000') // [!code ++]
+const host = values.host ?? 'localhost' // [!code ++]
+const verbose = values.verbose ?? false // [!code ++]
+```
+
+### yargs with subcommands
+
+`yargs` subcommand support cannot be directly replaced with `parseArgs`. You'll need to handle routing yourself:
+
+```ts
+// yargs approach // [!code --]
+yargs(hideBin(process.argv)) // [!code --]
+  .command('serve', 'Start the server', (yargs) => { // [!code --]
+    return yargs.option('port', { type: 'number' }) // [!code --]
+  }, (argv) => { // [!code --]
+    startServer(argv.port) // [!code --]
+  }) // [!code --]
+  .command('build', 'Build the project', {}, () => { // [!code --]
+    runBuild() // [!code --]
+  }) // [!code --]
+  .parse() // [!code --]
+
+// parseArgs approach // [!code ++]
+const { values, positionals } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    port: { type: 'string' }, // [!code ++]
+  }, // [!code ++]
+  allowPositionals: true, // [!code ++]
+}) // [!code ++]
+
+const [command] = positionals // [!code ++]
+switch (command) { // [!code ++]
+  case 'serve': // [!code ++]
+    startServer(Number(values.port)) // [!code ++]
+    break // [!code ++]
+  case 'build': // [!code ++]
+    runBuild() // [!code ++]
+    break // [!code ++]
+  default: // [!code ++]
+    console.error(`Unknown command: ${command}`) // [!code ++]
+    process.exit(1) // [!code ++]
+} // [!code ++]
+```
+
+> [!NOTE]
+> If your CLI relies heavily on yargs features like `.demandOption()`, `.conflicts()`, `.implies()`, auto-generated help with `--help`, or complex subcommand nesting, migrating to `parseArgs` requires implementing these features yourself. Evaluate whether the dependency savings justify the added code.
+
+## `commander` and `sade`
+
+`commander` and `sade` provide subcommand routing and auto-generated help, which `parseArgs` does not. For simple single-command CLIs, you can replace the parsing portion:
+
+```ts
+import { program } from 'commander' // [!code --]
+program // [!code --]
+  .option('-f, --force', 'Force operation') // [!code --]
+  .option('-o, --output <path>', 'Output path') // [!code --]
+  .parse() // [!code --]
+const opts = program.opts() // [!code --]
+
+import { parseArgs } from 'node:util' // [!code ++]
+const { values } = parseArgs({ // [!code ++]
+  args: process.argv.slice(2), // [!code ++]
+  options: { // [!code ++]
+    force: { type: 'boolean', short: 'f' }, // [!code ++]
+    output: { type: 'string', short: 'o' }, // [!code ++]
+  }, // [!code ++]
+}) // [!code ++]
+```
+
+> [!NOTE]
+> If you need subcommands, auto-generated help, or validation, `parseArgs` may not be sufficient on its own. Consider keeping `commander` or `sade` for complex CLIs, or build these features yourself.
+
+## Feature comparison
+
+| Feature | `parseArgs` | `minimist` | `yargs` | `commander` |
+|---------|-------------|------------|---------|-------------|
+| Boolean flags | ✅ | ✅ | ✅ | ✅ |
+| String options | ✅ | ✅ | ✅ | ✅ |
+| Short aliases | ✅ | ✅ | ✅ | ✅ |
+| Multiple values | ✅ `multiple: true` | ✅ | ✅ | ✅ |
+| Default values | ⚠️ manual `??` | ✅ | ✅ | ✅ |
+| Subcommands | ❌ | ❌ | ✅ | ✅ |
+| Auto-help | ❌ | ❌ | ✅ | ✅ |
+| Type coercion | ❌ (string/boolean only) | ❌ | ✅ | ✅ |
+| Validation | ❌ | ❌ | ✅ | ✅ |
+
+## Node.js version requirements
+
+`util.parseArgs` is available in:
+- Node.js 18.3.0+
+- Node.js 16.17.0+
+
+For older Node.js versions, use the [`@pkgjs/parseargs`](https://github.com/pkgjs/parseargs) polyfill.
+
+## Further reading
+
+- [Node.js util.parseArgs documentation](https://nodejs.org/api/util.html#utilparseargsconfig)
+- [parseArgs proposal and discussion](https://github.com/nodejs/node/pull/42675)