I build self-hosted infrastructure and desktop software for Linux, mostly in Rust and TypeScript. I care most about the unglamorous parts that make a system trustworthy: which side opens each connection, where every secret lives, what happens when a delivery is uncertain, and how an update proves where it came from.

A private workspace that connects an AI agent across several of my own computers without opening a single port on any of them.

• Computers dial out over TLS, and each has its own credential that can be revoked
• Passkey sign-in, and the hub credential never reaches the browser
• Signed, staged updates that are verified before they're activated atomically
• Journaled dispatch: prompt IDs are reserved before sending and never retried blindly
• End-to-end acceptance tests that start the real binaries, not mocks

The architecture, threat model, known limits and engineering notes are published at slaukokit.dev. The code is private for now.

Rust · TypeScript · Next.js · Tauri · SQLite · systemd · nftables · Linux

slaukokit.dev · SlaukoKit on GitHub