Tokens, Vectors, and Attention: How an LLM Actually Reads Your Prompt
An LLM never sees your words. It sees numbers: token IDs, vectors, and probabilities. Here's what actually happens between you hitting enter and the model's reply appearing, one step at a time.
26 July 2026
AI Liability: Who Is Responsible When AI Gets It Wrong?
MCP Explained: How AI Models Talk to Your Tools (and What Comes Next)
If You're Selling AI as a Cost Cutter, You're Selling It Wrong
Why Are You Paying So Much for AI Coding Tools?
Developer resources
Open-source agent skills, a CLI for managing them, and the design system that powers this site.
Install and share agent skills and MCP configs across Claude, Cursor, Windsurf, and other coding agents.
The skills this site runs on, plus 22 development harness skills with brownfield change control and verification loops.
28 research-backed UX laws, each with a without/with example, grounded in mechanism instead of taste.
Design system documentation with brand guidelines, fundamentals, UI components, and game design primitives.
Patterns, language features, and practical JS craft.
Layout, styling systems, and modern CSS techniques.
Protocols, systems, data architecture, and fundamentals beyond any one language.
Building, testing, and shipping AI-powered features.
Opinion pieces, industry takes, and things I am working through.
App Router, rendering, and full-stack React patterns.
Reliable test strategies for modern web applications.
Game engines, mechanics, and shipping interactive experiences.
Pathfinding Algorithms Explained
Tokens, Vectors, and Attention: How an LLM Actually Reads Your Prompt
SOLID Principles Matter More With AI Coding Agents, Not Less
AI Liability: Who Is Responsible When AI Gets It Wrong?
Building Minesweeper in React, Part 1: Client-Side Architecture
MCP Explained: How AI Models Talk to Your Tools (and What Comes Next)