The Work Can Look Done Before the Thinking Is Done
Why agent-assisted work can look finished before the thinking is finished, and why visible reasoning needs to become part of mentoring, review, and execution.
Transcripts over memory. Evidence over vibes.
Why agent-assisted work can look finished before the thinking is finished, and why visible reasoning needs to become part of mentoring, review, and execution.
How my Markdown notes grew into an interface for agents, and what that taught me about keeping useful friction while turning repeated pain into structure.
A small CLI tool for browsing, searching, and resuming copilot-cli sessions. Built on the SQLite database copilot-cli already maintains for you.
Your .env files are a liability. Two simple patterns for injecting secrets from 1Password or macOS Keychain so credentials never touch disk as plaintext.
Tri-State Relay Service
A free macOS relay for hearing short agent updates when you are ready for them.
Download and get startedA free macOS app and CLI for hearing short agent updates when you are ready for them, without keeping every agent session open.
A framework for improving AI agent development through systematic learning from failures. Design environments where the easiest path is the correct one.
A resource for mastering AI coding assistants, helping developers configure and optimize their AI agent tools.
I write here because writing is how I think.
Most posts come from real projects: what I tried, what worked, what surprised me. If a post or project helps you solve a problem at work tonight, that is the best outcome I can imagine.