Skip to content

Blog

Notes on agentic ops, safely.

Writing in progress. Here’s what’s coming first.

Coming soon

Why we let our AI agent run on a local model — and you should too

The compliance case for local-first agentic ops.

Bring-your-own-key AI tools: who actually sees your data?

A clear-eyed look at where your tokens go.

A safer way to give an AI shell access: confirm, gate, audit

The design behind the three safety modes.

Incident triage with an agentic SSH workflow

A worked example, transcript included.