kanzo
Self-hosted kanban board and MCP server for humans and coding agents.
Single binary browser board with built-in MCP server for AI coding assistants.
Kanzo is a 4.7 MB self-hosted kanban board that integrates directly with coding agents through the Model Context Protocol. It is built for developers who want a shared task space where both they and their AI agents can manage work.
What it is
It runs a local browser board and MCP server simultaneously, allowing agents to pick tasks, move cards, and comment on progress. Features include local-first storage, task dependencies, filtering, and optional P2P sharing or git-backed sync.
Who it's for
It suits developers working alongside AI coding assistants like Claude Code or Codex who need shared task tracking. It does not suit teams looking for a traditional cloud-hosted enterprise project management suite.
Key features
Built-in MCP Server
Lets coding agents fetch tasks, move cards, and leave comments directly.
Zero Setup Binary
Runs instantly from a single lightweight binary using 5 MB of RAM.
Task Dependencies
Blocks tasks from being started until prerequisite tickets are marked done.
Local-First Sync
Keeps your board data in a local folder with optional private git repo syncing.
Local Network Sharing
Shares the board across multiple computers on your network without configuration.
Up and running in 3 steps
Our honest verdict
What's great
- Extremely lightweight with a 4.7 MB binary and 5 MB RAM footprint.
- Native MCP integration allows AI agents to autonomously manage workflow states.
- Supports local network sharing between multiple machines instantly.
- Includes a git-backed sync option for moving between devices.
Where it falls short
- Requires manual terminal configuration to set up MCP clients like Claude Code.
- Local network sharing has no built-in authentication, so you must trust your network.
- Running it continuously as a service on a Raspberry Pi requires manual setup.
Pricing & license
Free under its AGPL-3.0 licence, with the only cost being the hardware and electricity required to run it yourself.
Open remote AI-training roles
767 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
How do coding agents connect to Kanzo?
You add Kanzo as an MCP server using your client's configuration command, allowing the agent to interact with the running board.
Can multiple people use the same board?
Yes, you can turn on board sharing and have other computers on your network join the address.
Where is board data stored?
Your entire board lives in a local `~/kanzo` directory created on the first run.
