AI-Devkit (github.com)

AI-Devkit is a CLI toolkit that scaffolds AI-assisted software development by generating phase-based Markdown templates (with YAML frontmatter) and automatically configuring workspaces for Cursor and Claude Code. You can run it via npx or install globally; the interactive ai-devkit init (or flags like --environment cursor, --all, --phases ...) creates a .ai-devkit.json state file, a docs/ai/ folder with staged READMEs (requirements, design, planning, implementation, testing, deployment, monitoring), and environment-specific folders (.cursor/ and .claude/) containing project rules and slash/command files. Templates encourage mermaid diagrams for architecture/data flow and are fully customizable; the CLI also supports adding/updating individual phases (ai-devkit phase testing). Technically significant for AI/ML teams, AI-Devkit standardizes an AI-friendly development lifecycle and embeds guided workflows into AI pair-programming tools. Slash commands (e.g., /new-requirement, /execute-plan, /code-review, /writing-test, /capture-knowledge) surface directly in Cursor and Claude Code chats to drive tasks like end-to-end feature creation, structured local code reviews, automated test generation, and deep codebase analysis. capture-knowledge can recurse dependencies, emit mermaid diagrams, and save outputs (options: --depth, --save, --diagram-only), enabling reproducible documentation, onboarding, and refactoring with visual artifacts versioned alongside code. The tool also auto-initializes git if missing, is MIT-licensed, and is aimed at making AI assistants more contextual, auditable, and aligned with project processes.