agentic-development

What problem does it solve?

Traditional software development methodologies often hinder the efficiency of AI agents, leading to over-engineering and frustration. This Skill provides a pragmatic, "just talk to it" approach to agentic development, optimizing for speed, iteration, and developer well-being.

Core Features & Use Cases

• Efficient Agent Workflows: Guides on parallel agent management, blast radius thinking, and interrupting agents for active steering, maximizing velocity.
• Optimized Prompting: Advises on model-specific prompt strategies, emphasizing shorter prompts for capable models and the power of screenshots for visual context.
• Tool & Model Selection: Offers insights into model economics, characteristics (GPT-5-Codex vs. Claude Sonnet), and the benefits of CLIs over complex MCPs.
• Refactoring & Testing: Recommends using agents for low-focus refactoring tasks and generating tests in the same context to catch immediate bugs.
• Use Case: You're struggling with managing multiple AI agents on a large codebase. Use this skill to learn how to run parallel agents in one folder, effectively use screenshots for context, and refactor code efficiently without burning out.

Quick Start

I'm starting a new project with AI agents. How should I set up my workflow for maximum efficiency, and which model should I choose for coding tasks?