AI that still helps next Tuesday.
Notes on agents, voice capture, knowledge systems, and the habits that keep them useful once real life gets involved.
Recent
- 3 min readGitHub 101 for small teams who just want to stay in sync
A plain-English guide for non-developers using GitHub as a shared, versioned knowledge home.
- 8 min readDocumentation just got 10x more valuable
A take on the knowledge paradox: the reader of your documentation changed, so capturing it is suddenly worth far more than tidying it. Talk, transcribe, and spend your time pointing at the work rather than filing it.
- 4 min readI packaged my agent setup into a playbook you can run
A build log of agents-first-playbook — what is in the repo, how the read-only inspector works, and what you can lift for your own multi-agent setup.
- 5 min readWhat an MCP server actually is
A plain explainer of MCP servers, told through two I have worked on — one that talks straight to an API, and one that is only a thin proxy.
- 7 min readYour AI instructions shouldn’t belong to one tool
A practical note on migrating from a mature Claude-first setup to an AGENTS-first setup without breaking the working system.
- 2 min readMarkdown was supposed to be the simple bit
A short note on .md, .mdx, and the irony of discovering that even simple document formats have edges.
- 2 min readThe voice note was not the hard part
A personal note on why capture only starts to matter when it becomes operating context.