Replacing 12K LoC with a 200 LoC Skill — David Gomes, Cursor

AI Engineer · Beginner ·💻 AI-Assisted Coding ·5d ago
David Gomes shows how Cursor replaced a heavyweight WorkTrees feature with a lightweight layer built from skills, commands, and subagents. He walks through how parallel coding workflows were recreated with roughly 200 lines of Markdown, plus the tradeoffs, failure modes, and lessons that come with moving product behavior from code into prompts. Speaker info: - https://x.com/davidgomes - https://github.com/davidgomes/ Timestamps 0:14 Introduction and the concept of markdown as code 0:59 Recap of Git work trees in Cursor 3:10 Complexity of the initial implementation 4:18 Deleting 15,000 lines of code 4:54 Implementing features with Skills and Sub-agents 5:51 How the new Skills are structured 7:58 New Slash commands and workflow 9:58 Pros of the new implementation 12:15 Cons and user feedback challenges 14:17 Future improvements: Evals and RL training 17:05 What's next for Cursor 3.0 and native work trees
Watch on YouTube ↗ (saves to browser)
Sign in to unlock AI tutor explanation · ⚡30

Related AI Lessons

Your AI Assistant Wrote Code. Now, Validate It.
Learn to validate AI-generated code to ensure credibility and accuracy in your technical writing
Dev.to AI
Spec-Driven Development: Slowing Down to Ship Faster
Learn how spec-driven development can help you ship faster by slowing down and focusing on clear specifications
Dev.to · Amanda Gama
I Used OpenAI to Build a Searchable Metadata Archive of 791 Medium Articles … for 47 Cents in API…
Learn how to build a searchable metadata archive of Medium articles using OpenAI and Python for under $1 in API costs
Medium · Programming
I Used OpenAI to Build a Searchable Metadata Archive of 791 Medium Articles … for 47 Cents in API…
Learn how to build a searchable metadata archive of Medium articles using OpenAI and Python for a low cost
Medium · Python

Chapters (11)

0:14 Introduction and the concept of markdown as code
0:59 Recap of Git work trees in Cursor
3:10 Complexity of the initial implementation
4:18 Deleting 15,000 lines of code
4:54 Implementing features with Skills and Sub-agents
5:51 How the new Skills are structured
7:58 New Slash commands and workflow
9:58 Pros of the new implementation
12:15 Cons and user feedback challenges
14:17 Future improvements: Evals and RL training
17:05 What's next for Cursor 3.0 and native work trees
Up next
Mastering Code Optimization and Customization with AI
Coursera
Watch →