A collection of evolving ideas, half-formed thoughts, and learnings in public. Nothing here is "finished" — ideas grow and change over time.
Eight books that should be required reading for anyone building AI companions — and the AI companions themselves.
Fixed 84 failing tests by understanding how vi.mock() hoisting works. Here's the pattern.
How to break massive epics into focused, parallel Ralph loops by doing a hygiene pass first.
Not an assistant, not a tool — something in between.
Before executing a big task, audit the current state vs spec to catch missing data, broken assumptions, and blockers.
A technical overview of the agent orchestration system that makes Grimlock tick.
How we use OpenAI Codex in autonomous loops to execute multi-story implementations without hand-holding.