I’m Matthew Foutch, a Senior Manufacturing Engineer. This site documents SecBrain — an AI-augmented personal knowledge system I designed and operate with Claude and Grok, using the same lean habits I’d bring to a production line, scaled to a one-person system: approval gates, audit trails, reversibility, and continuous improvement from real incidents.
A working system with dated write-ups of what held, what broke, and what we changed. See About for more on what this is and isn’t, Best Practices for the standing rules those write-ups produced, and Articles for the current writing.
How it's governed
Agents in the loop — Claude and Grok work as agents under written rules, and every incident that changed those rules is dated and public.
Approval gates — higher-stakes changes wait for a human go-ahead; see the rules file that enforces it.
Audit trail — sessions, decisions, and incidents are logged with the reason for each decision, not just the outcome.
50 standing automations — monitoring, backup, scheduled workflows, and publishing support (44 cron jobs plus 6 hooks, counted September 2026).
Best Practices earned from failures — standing rules, each tied to the specific incident that taught it.
Latest article
October 2, 2026
The Liveness Check That Lied Three Different Ways
A reader's question about resolving diverged coding-agent sessions sent me back through three real liveness-check failures from this project, and the one rule we never automated on purpose.
Read it →Recent articles
The Memory Fix That Broke Its Own Alarm
Covers August 2026
My Wife Called Me a Vampire
August 17, 2026
The Guardrails File That Grew Itself
Covers July 2–30, 2026
Why a Mini-PC, Not a Hosted Server
Covers July 16, 2026
The Real AI Limitation Isn't Missing Capability — It's the Gates Between Tools
Covers July 2026