Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate Yargs-based CLI applications with commands, positional args, middleware, and TypeScript support. Creates a complete scaffolded CLI application with modern patterns.
.claude/skills/a5c-ai-yargs-scaffolder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 1% | 0% |
Generate a complete Yargs CLI application with TypeScript, middleware support, and best practices.
Invoke this skill when you need to:
| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | projectName | string | Yes | Name of the CLI project (kebab-case) | | description | string | Yes | Short description of the CLI | | commands | array | No | List of commands to scaffold | | typescript | boolean | No | Use TypeScript (default: true) | | packageManager | string | No | npm, yarn, or pnpm (default: npm) | | strictMode | boolean | No | Enable strict mode (default: true) |
json{ "commands": [ { "name": "serve", "description": "Start the server", "aliases": ["s"], "positional": [ { "name": "port", "type": "number", "default": 3000 } ], "options": [ { "name": "host", "type": "string", "default": "localhost" }, { "name": "watch", "type": "boolean", "alias": "w" } ] } ] }
<projectName>/
├── package.json
├── tsconfig.json
├── .gitignore
├── README.md
├── src/
│ ├── index.ts # Entry point
│ ├── cli.ts # Yargs setup
│ ├── commands/
│ │ ├── index.ts # Command exports
│ │ └── <command>.ts # Command modules
│ ├── middleware/
│ │ ├── logger.ts # Logging middleware
│ │ └── config.ts # Config loading middleware
│ ├── utils/
│ │ └── helpers.ts # Helper utilities
│ └── types/
│ └── index.ts # Type definitions
└── tests/
└── commands/
└── <command>.test.tstypescriptimport yargs from 'yargs'; import { hideBin } from 'yargs/helpers'; import * as commands from './commands'; import { loggerMiddleware } from './middleware/logger'; export const cli = yargs(hideBin(process.argv)) .scriptName('<projectName>') .usage('$0 <cmd> [args]') .middleware([loggerMiddleware]) .command(commands.serveCommand) .command(commands.buildCommand) .demandCommand(1, 'You need at least one command') .strict() .fail((msg, err, yargs) => { if (err) throw err; console.error(msg); console.error(yargs.help()); process.exit(1); }) .help() .alias('help', 'h') .version() .alias('version', 'v') .wrap(Math.min(120, process.stdout.columns || 80));
typescriptimport { CommandModule, Argv } from 'yargs'; interface ServeArgs { port: number; host: string; watch: boolean; } export const serveCommand: CommandModule<{}, ServeArgs> = { command: 'serve [port]', aliases: ['s'], describe: 'Start the development server', builder: (yargs: Argv) => { return yargs .positional('port', { type: 'number', default: 3000, describe: 'Port to listen on' }) .option('host', { type: 'string', default: 'localhost', describe: 'Host to bind to' }) .option('watch', { alias: 'w', type: 'boolean', default: false, describe: 'Enable watch mode' }); }, handler: async (argv) => { console.log(`Starting server on ${argv.host}:${argv.port}`); } };
json{ "dependencies": { "yargs": "^17.0.0" }, "devDependencies": { "@types/node": "^20.0.0", "@types/yargs": "^17.0.0", "typescript": "^5.0.0", "tsx": "^4.0.0", "vitest": "^1.0.0" } }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 18,949 | 16,567 | -13% | 1 | 1 | 0% | 4,352 | 5,425 | +25% | 0 | 0 | — |
case-02 | fail→pass | 14,955 | 14,040 | -6% | 1 | 1 | 0% | 3,584 | 4,782 | +33% | 0 | 0 | — |
case-03 | fail→pass | 17,007 | 17,098 | +1% | 1 | 1 | 0% | 3,948 | 5,619 | +42% | 0 | 0 | — |
case-04 | fail→fail | 11,918 | 8,346 | -30% | 1 | 1 | 0% | 2,439 | 3,105 | +27% | 0 | 0 | — |
case-05 | fail→fail | 16,855 | 16,823 | -0% | 1 | 1 | 0% | 3,795 | 4,390 | +16% | 0 | 0 | — |
case-06 | fail→fail | 10,717 | 9,081 | -15% | 1 | 1 | 0% | 2,283 | 3,361 | +47% | 0 | 0 | — |
case-07 | fail→fail | 8,102 | 4,113 | -49% | 1 | 1 | 0% | 1,527 | 2,226 | +46% | 0 | 0 | — |
case-08 | fail→fail | 8,464 | 9,309 | +10% | 1 | 1 | 0% | 1,774 | 3,338 | +88% | 0 | 0 | — |
case-09 | fail→fail | 9,386 | 6,615 | -30% | 1 | 1 | 0% | 2,136 | 2,986 | +40% | 0 | 0 | — |
case-10 | fail→fail | 11,863 | 8,139 | -31% | 1 | 1 | 0% | 1,980 | 2,703 | +37% | 0 | 0 | — |
case-11 | fail→pass | 9,550 | 5,391 | -44% | 1 | 1 | 0% | 1,852 | 2,486 | +34% | 0 | 0 | — |
case-12 | fail→pass | 10,717 | 3,975 | -63% | 1 | 1 | 0% | 2,340 | 2,375 | +1% | 0 | 0 | — |
case-13 | fail→fail | 9,938 | 7,228 | -27% | 1 | 1 | 0% | 2,171 | 2,978 | +37% | 0 | 0 | — |
case-14 | fail→pass | 9,119 | 5,671 | -38% | 1 | 1 | 0% | 1,724 | 2,563 | +49% | 0 | 0 | — |
case-15 | fail→fail | 5,615 | 5,440 | -3% | 1 | 1 | 0% | 1,232 | 2,568 | +108% | 0 | 0 | — |
case-16 | fail→fail | 11,909 | 6,200 | -48% | 1 | 1 | 0% | 2,140 | 2,686 | +26% | 0 | 0 | — |
case-17 | fail→fail | 9,558 | 2,753 | -71% | 1 | 1 | 0% | 1,433 | 1,939 | +35% | 0 | 0 | — |
case-18 | fail→fail | 14,546 | 11,182 | -23% | 1 | 1 | 0% | 2,342 | 3,228 | +38% | 0 | 0 | — |
case-19 | fail→fail | 7,086 | 5,145 | -27% | 1 | 1 | 0% | 1,484 | 2,343 | +58% | 0 | 0 | — |
case-20 | fail→pass | 11,033 | 6,829 | -38% | 1 | 1 | 0% | 1,979 | 2,566 | +30% | 0 | 0 | — |
case-21 | fail→fail | 10,603 | 5,399 | -49% | 1 | 1 | 0% | 1,906 | 2,277 | +19% | 0 | 0 | — |
case-22 | fail→fail | 9,307 | 4,181 | -55% | 1 | 1 | 0% | 1,370 | 2,229 | +63% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.