Hey Product Hunt! 👋 I'm the maker of Maestro. Here's the story: I spent months watching AI coding agents make the same workflow mistakes — hallucinating APIs, skipping error handling, building multi-agent systems when a single agent would do, and claiming "Done!" without running a single test. The code wasn't the problem. The workflow was. So I built Maestro — a skill system that gives any AI agent structured workflow discipline. Think of it as a playbook: 21 commands that cover the full lifecycle from diagnosis to hardening to final polish. What makes it different: It's not another AI tool. It's instructions for your existing AI tools. Works with Claude, Cursor, Copilot, VS Code, and 7 more. Anti-patterns are first-class. Every command explicitly tells the AI what NOT to do. This alone eliminates most workflow failures. Commands chain. /diagnose → /fortify → /evaluate → /refine. Each step recommends the next. It's fully open source (MIT), available as skills files, an MCP server, and a VS Code extension. I'd love your honest feedback — what commands would you add? What workflows break your AI agents the most? 🔗 https://github.com/sharpdeveye/m... | Try it

Maestro

Workflow fluency for AI coding agents

Workflow fluency for AI coding agents