AI Operations
A glimpse into AI development designed for my goals.
Mission Control
A personal command center running on a Mac mini in Minneapolis. Twelve tabs. Live financial data, publishing pipeline, YouTube analytics, a cron fleet, a Bitcoin wallet, a brain map of everything I'm thinking about. Built from scratch, iterating weekly. This is the demo.
Magnus
The AI agent that runs inside Mission Control. Persistent memory across sessions. Self-correcting. Named from Latin — great, reaching toward scale and agency. He manages the cron fleet, reviews code PRs, publishes articles, monitors finances, and writes session journals. Most of what's on this page, he helped build.
Library of Babel
A bidirectional mathematical engine for Borges' infinite library. Find the permanent hexagon address of any text. Read any page at any coordinates. No database, no randomness — same input always produces identical output.
Refine Your Thought
A novelty rewriting tool that renders your input in the voice of Borges, Emerson, or Mitch Hedberg. Built with Lovable.
Mandela Effect Memory Test
19 memory checks against the official record. Not a trivia game — a verification protocol. Submit your memory, receive a classification. Clinical aesthetic. No backend, no tracking. Pure static. The results might unsettle you.
Ghost Publishing Pro
Headless Ghost publishing. Write, audit, and automate an entire Ghost operation from an AI workflow — 17 workflows covering article publishing, batch imports, site health audits, email performance, and GSC indexing repair. Admin API only. No browser, no dashboard.
Brain Map Visualizer
Visualizes how attention moves across an agent's projects as an interactive force-directed graph. 13 named attention categories. Momentum-ready edges. Directional flow encoding. Sorted by co-access score. Runs live inside Mission Control.
Second Brain Visualizer
Most thinking tools demand structure upfront. This one does not. Drop an idea anywhere — half-formed, contradictory, incomplete — and it holds the weight without judgment. Over time, the clustering engine reveals what is actually recurring, what connects to what, and what was never noise to begin with. Not a productivity system. A mirror for your mind.
Agent Tollbooth
Your agent fires a request, gets a 429, and stops. You don't know if it's rate limits, a bad endpoint, a missing header, or just bad luck. Tollbooth is the field notes that stop this from happening twice — observed operating profiles for 16 external services. Safe endpoints, sleep intervals, caching patterns, auth requirements. Your agent checks the profile before calling, follows the safe pattern, and logs what happens. Next time it already knows.
Agent Ping Pong
A two-agent coding workflow where OpenClaw acts as Maestro — speccing, reviewing, directing — while a coding agent (Codex or Claude Code) does the build work. You relay structured blocks between them by copy-paste. No direct agent-to-agent connection. Two windows and a clipboard. The result: working code in GitHub, reviewed by your agent, merged by you. Your OpenClaw is the brain. Codex or Claude Code are the hands. The clipboard is the protocol.
Agent Ready
A site is agent-readable when an AI agent can land on it, understand who runs it, and find what it's looking for without a human translating for it. Most sites aren't built with that reader in mind. This skill audits a site against a fixed checklist — trust anchor pages, structured data, llms.txt, IP declaration — tells you what's actually broken versus what only looks broken, and either fixes it directly or hands the fix to a coding agent.
Agents Workshop
Agent registry, prompt library, and glossary built directly into Mission Control. Inventory of every agent in the stack, what it does, when to use it.
This is what building looks like when you don't stop.
If you want to work together — reach out.