AI Development with ShipClojure

ShipClojure comes with built-in support for AI-assisted development through various LLM tools. This guide explains how to leverage AI tools effectively to accelerate your development workflow.

Getting Started with AI Tools

ShipClojure includes pre-configured documentation and context files that help AI tools understand the codebase structure, conventions, and best practices. These files make AI tools like Claude Code, Cursor, Windsurf, and others more effective at generating quality code that matches the project's patterns.

Setting Up Your AI Tool

1. Copy the CLAUDE.md file:

   • The primary AI context file is located at /docs/pragma/development/docs/llms/CLAUDE/
   • Copy this file into your AI editor's context or reference it directly
   • This file contains essential information about the project's structure, conventions, and patterns

2. For Claude Code users:

   • Claude Code will automatically detect the CLAUDE.md file in your project

3. For Cursor users:

   • Create a chat and reference the CLAUDE.md file
   • Use /docs/pragma/development/file docs/llms/CLAUDE/ to include it in your context

4. For other AI coding assistants:

   • Either paste the content of CLAUDE.md at the beginning of your conversation
   • Or point the tool to the file path for reference

Effective AI Development Strategies

Keep Tasks Focused

Break down your development tasks into focused chunks for better AI assistance:

• Build UI first, add interactivity later:

  • Ask AI to build a UI page or component structure first
  • In a separate task, add event handlers and state management

• Component development:

  • Request AI to create a basic component
  • Once complete, ask for component scenes for documentation
  • Finally, integrate the component into your application

• Testing assistance:

  • Ask AI to write tests for specific functions or components
  • For complex cases, provide example tests for similar components

Context-Aware Help

ShipClojure comes with bundled documentation for:

• UIx (React wrapper)
• DaisyUI components
• Re-frame integration
• Database interactions
• Authentication flows
• WebSocket communication

This built-in context makes LLMs particularly proficient at:

1. Creating new UI components following project conventions
2. Implementing Re-frame events and subscriptions
3. Writing API routes with proper schema validation
4. Creating database migrations and queries

Debugging with AI

When encountering issues:

1. Share the error message with the AI
2. Include relevant code snippets
3. Ask the AI to explain what might be causing the issue
4. Request potential solutions to try

Limitations and Best Practices

• Split complex features: Break large features into smaller, focused tasks
• Verify generated code: Always review and test AI-generated code
• Provide examples: When available, show examples of similar code from the project
• Iterative improvement: Use AI to refine and improve code in multiple passes

Remember that AI tools work best as collaborative partners. They excel at generating boilerplate, suggesting approaches, and implementing patterns that already exist in your codebase.