MIKE

Feature Highlights

🗂️

Parallel work streams

It's the familiar kanban you know (and hopefully love). Each card represents a different task, and you can run as many as your system (or inference provider) can handle.

🔒

Strong sandbox isolation

All agent activity runs inside gVisor-isolated containers via rootless Podman, so they only have access to what you allow. Point-and-click configuration makes it easy set up and review.

🌿

Worktrees & Threads

Git worktrees are used in combination with sandboxes and a read-only repository mount to provision agents securely and efficiently. Threads makes it easy to quickly explore different ideas.

📱

Work from anywhere

This is a bit of a double-edged sword, and hopefully the ability to work from anywhere manages to improve your work/life balance. And unlike terminal interfaces, MIKE is still very usable on high latency networks.

🔌

Secure extensions

Extend MIKE with custom extensions that run inside gVisor, iframe, and/or WASM sandboxes, helping ensure third-party features stay safely fenced in.

🦙

Bring your own inference

Built for llama.cpp, and works with any provider exposing the OpenAI Chat Completions API. Local-first by design.

Hello, world

October 10, 2026

Hi, I'm Mike. I've been building intranet tools in a large organization for about 20 years, and like many other developers, trying to adapt to changes in the new status quo.

This project aims to provide a simple and secure interface to facilitate code review, quick modifications, and delegation of tasks to agents.

I hope you find this useful.

Backlog

In Progress

Review

Done