Writing
Field notes on AI-native building, healthcare systems, and operational craft.
The Layers Behind One Website Revision: How My AI Agent Stack Actually Runs
A case study of one incremental website revision moving through planning, orchestration, specialist agents, memory, tests, review, and delivery.
Google Secret Manager: Stop Hardcoding API Keys
Google Secret Manager moves hardcoded API keys out of source code and into versioned, IAM-controlled storage. Follow a safer migration and rotation path.
How I Got Opus Back Under My Max Plan
How I run Claude Opus 5 in my own coding setup using the Max subscription I already pay for, what broke first, and where the limits are.
How I Found My Voice in 13,839 Messages
I built a writing profile from the OpenCode messages I actually typed, with clear limits, an authorship gate, and a test against unseen work.
Day Zero: I Audited How My AI Projects Actually Begin
A public field note from auditing 94 AI project directories and turning project starts into a clearer Day Zero ritual.
How I Run AI Agents: Fable for Planning, a Slim Harness for Execution
A public field note on using Fable for plan mode, GPT-5.5 high for bounded execution, and oh-my-opencode-slim to keep agent work disciplined.
Commitmaxxing & Tokenminimizing: The New Discipline for AI-Native Builders
A practical field note on shipping with AI agents by making every useful step durable and every context window smaller.