Quick Start

Get a board running in under 60 seconds. No configuration required.

1. Install globally

   npm install -g kanban-lite

   Or skip the global install and use npx for every command:

   npx kanban-lite init
   npx kanban-lite serve

2. Initialize your board

   cd my-project
   kl init

   Creates a .kanban/ directory with an empty default board in your current project directory.

3. Start the web UI

   kl serve
   # → Board running at http://localhost:3000

   Opens the board at localhost:3000. All connected clients receive real-time updates via WebSocket.

4. Create your first card

   kl add --title "My first task" --priority high

   The card appears in the web UI immediately. By default it lands in the Backlog column.

5. Use the CLI to manage cards

   kl list                           # see all cards
   kl list --status todo --priority high  # filter by status & priority
   kl move my-first-task in-progress # move to In Progress
   kl edit my-first-task --assignee alice --due 2026-04-01

6. For AI agents — start the MCP server

   kl mcp

   Exposes all board operations as MCP tools. Works with Claude, Codex, and any MCP-compatible agent.

What just happened?

kl init created a .kanban/ directory. Inside that directory, each kanban card is a plain .md file with YAML frontmatter, organized into sub-directories by status. Cards are human-readable, version-controllable, and diff-friendly — just like source code.

Commit .kanban/ to your git repository. Every card change becomes a commit, and any team member can pull and see the current board state.

Next steps

If you want runnable integration walkthroughs instead of a blank board, head to the examples hub for the shipped Chat SDK / Vercel AI, LangGraph Python, and Mastra Agent Ops guides.