Reward-Based Self-Improvement for LLMs
A pattern for improving AI output quality by ~15% using incentive framing in system prompts. Treat your LLM like a game character.
Jan 1, 2026ai, llm
Thoughts on design, development, and building things
A pattern for improving AI output quality by ~15% using incentive framing in system prompts. Treat your LLM like a game character.
How to build autonomous testing suites that improve themselves using AI consensus, clear completion criteria, and parallel agents.
Generate consistent, accessible color scales from any base color in seconds.
Create harmonious type scales using mathematical ratios for consistent, beautiful typography.
A collection of skills that give Claude Code new capabilities for design systems, accessibility, and development.
Generate mathematical spacing scales that bring harmony to your layouts.