GitHub Spec Kit makes AI agents write a spec before coding
GitHub's open-source Spec Kit uses commands like constitution, specify, plan and tasks to force agents to produce a structured specification before touching code, and works with Claude Code, Cursor, Codex and more than 25 other agents.
THIS IS F**CKING INSANE GitHub has just solved the biggest problem with vibe coding. They've just launched Spec Kit and in a short time it already has +126K stars. The idea? Instead of throwing vague prompts and praying the agent doesn't break your project. Spec Kit forces the AI to create a structured specification BEFORE touching code. The AI first understands what you want to build, asks about what's missing, organizes the project, and then starts programming. That means less time fixing absurd errors, less inconsistent code, and much more predictable results when working with agents. The flow is simple: /constitution → rules and standards /specify → what you want to build /clarity → doubts before starting / plan → architecture and stack /tasks → ordered tasks /implement → execution Compatible with Claude Code, Cursor, Copilot, Codex, Gemini CLI, and +25 agents. 95K stars. 8K forks. Open source. Published by GitHub. Link to the repository 👇