Lab
Experiments and working tools from the practice — built with the same systems discipline we sell, published so you can inspect them. Everything here describes capabilities and implementation patterns, not unmeasured outcome claims.
Interactive map
AI Coding Agent Tech Tree
Thirty agentic-coding capabilities across five tiers — from Claude Code fundamentals through hooks, skills, MCP servers and subagents up to agent swarms. Each node: what it is, why it matters, how to implement it, and the pitfalls.
Open source
The toolshelf on GitHub
Internal practice builds moved into public: prospecting radars, outbound infrastructure, creative engines and GEO tooling. Next out of the queue: page-auditor, a SEO/GEO page grader.
Working tool
Site readability auditor
The page-grading tool that used to front this site: it reads a page the way AI engines do and ranks the fixes. Still runs on request; the build notes and weekly findings live on the Writing page. Ask on a call if you want your site graded.