Quickstart

Last updated February 20, 2026

Get AutoMD running locally in under 5 minutes. By the end, you’ll have a kanban board powered by a plain markdown file.

Prerequisites

  • Docker installed and running (get Docker)
  • A terminal (any OS)

Start AutoMD

Pull and run the AutoMD container:

docker run -d \
  --name automd \
  -p 3000:3000 \
  -v automd-data:/data \
  ghcr.io/automd/automd:latest

Open http://localhost:3000 in your browser. That’s it — AutoMD is running.

Create your first board

AutoMD turns any markdown file into a kanban board. Create a file called tasks.md:

# Backlog

- [ ] Design landing page @alice #design priority:high
- [ ] Set up CI pipeline @bob #devops est:4h

# In Progress

- [ ] Build auth module @charlie #backend due:friday

# Done

- [x] Write project brief @alice #planning

Drop this file into your AutoMD data directory, or paste it into the editor view. You’ll immediately see:

  • Editor view — your raw markdown with syntax highlighting
  • Checklist view — a filterable task list
  • Kanban view — columns from your # Headings, cards from your - [ ] items

Understanding the syntax

Each task is a standard markdown checkbox with optional metadata tags:

TagExamplePurpose
@name@aliceAssign to a person
#label#designCategorize with labels
priority:priority:highSet priority (high, medium, low)
due:due:fridaySet a due date
est:est:4hEstimate effort

It's just markdown

Everything is stored as plain text. You can edit files with any editor, commit them to git, or let AI agents manage them via MCP.

Connect an AI agent

AutoMD exposes an MCP (Model Context Protocol) server with 20+ tools. Add it to your AI agent’s config:

{
  "mcpServers": {
    "automd": {
      "url": "http://localhost:3000/mcp"
    }
  }
}

Now your AI agent can create tasks, move cards, update priorities, and more — all by writing to the same markdown files.

Info

See the full MCP Integration guide for detailed setup instructions per agent (Claude, Copilot, Cursor, etc.).

Next steps

Installation All deployment options: Docker, Docker Compose, and building from source. Core Concepts Understand boards, tasks, metadata, and the three views. Metadata Syntax Full reference for @assignee, #labels, priority, due dates, and more.