cli-generator_skill

Overview

This skill generates AI-friendly Python command-line interfaces using Click, Pydantic, Rich, and uv. It scaffolds a project layout, Pydantic response models, and conversational terminal output so each CLI interaction becomes a useful prompt for an agent. The output emphasizes actionable next steps and structured responses for agentic coding environments.

How this skill works

The generator creates a project skeleton (pyproject.toml, src package, commands, models, output, and core modules) and adds Click-based command entry points with epilog examples. It generates Pydantic models for structured CommandResult and ErrorDetail payloads and a ConversationalOutput class that prints success, error, and table views following the "Every Output is a Prompt" patterns. Templates include success suggestions, three-part error messages (what/how/next), and limited table rendering to avoid overwhelming agents.

When to use it

• Starting a new Python CLI that will be used by AI agents or autonomous workflows
• Standardizing CLI outputs so downstream agents can parse and act on results
• Building tooling that must suggest exact next-commands after each action
• Creating CLI commands with validated, machine-readable responses
• Adding friendly, example-rich --help epilogs for discoverability

Best practices

• Treat every CLI output as a prompt: include context, structured data, and suggested next commands
• Use Pydantic models to validate command responses and errors before rendering
• Follow the three-part error pattern: what went wrong, how to fix, and what's next
• Limit table columns and rows to keep agent inputs concise and scannable
• Include concrete examples in command epilog for reproducible agent steps

Example use cases

• Generate a search command that returns CommandResult with resource IDs and suggestions for view/export
• Create an auth flow command that prints step-by-step fixes and follow-up auth test commands on failure
• Scaffold export commands that show available formats and exact commands to run next
• Produce a tool for batch jobs where each job output includes suggested retry or debug commands
• Build a developer-facing CLI that agents can inspect and use to chain multi-step workflows