Lecture on AI Coding Tools and Essential Directories

Key Concepts

• AI Coding Tools: Tools that understand and work with your codebase. Examples include Cursor, Windsor, Fline, Codeex, and Claude code.
• Context is King: The importance of context in AI-based coding; without it, the AI cannot perform efficiently.

Essential Directories

1. AI Docs

   • Acts as persistent memory for AI coding tools.
   • Contains documentation such as third-party API documentation, integration details, custom patterns, and conventions.
   • Used primarily for third-party documentation to quickly ramp up codebases.

2. Specs Directory

   • Also known as specifications or plans (e.g., PRDs).
   • Critical for massive work done with AI tools.
   • Allows for agentic coding which uses multiple tools inside a single prompt.
   • Essential for detailed planning to hand off to AI tools.

3. Dotcloud (.cloud)

   • Specific to Claude code but applicable across other AI tools.
   • Contains reusable prompts for agentic coding tools.
   • Important for setting up new instances of agentic tooling efficiently.
   • Context priming prompts are crucial for setting up the initial context for AI tools.

Agentic Coding

• A superset of AI coding that involves creating self-validating loops in prompts.
• Encourages writing comprehensive plans and handing them to AI tools for execution.

Practical Example: Implementing a Feature

• Feature: Update the 'add' command in Pocket Pick to allow ID input.
• Workflow:
  1. Context Prime: Initial setup using context priming prompts for efficient operation.
  2. Plan Drafting: Creating a draft plan with the AI tool and iterating on it.
  3. Execution: AI tool executes the plan, making necessary changes and validating them.

Best Practices

• Use reusable prompts to save time and increase efficiency.
• Always prepare a detailed plan before execution.
• Allow AI tools to draft plans and then iterate on them.
• Shift from iterative prompting to creating comprehensive, agentic plans.

Conclusion

• The focus should be on patterns and principles, not specific tools.
• Essential directories help in scaling work across codebases.
• Great planning leads to great prompting, which allows efficient use of AI tools.
• AI Docs, Specs, and Dotcloud directories form the foundation for scaling engineering work with AI tools.

Additional Insights

• Importance of the context window in AI tools and managing it effectively.
• Emphasis on not iteratively updating codebase to avoid bad states.
• The role of AI tools in transforming engineers into curators of information.

Keep building and stay focused on implementing these strategies to optimize your engineering workflow with AI tools.