kanopy
Self-hosted kanban board where AI agents and people share projects and tasks
kanopy is an open-source MCP server for productivity. It can run locally on your machine or as a hosted remote server.
- Pricing
- Open source
- Platforms
- Web, API, CLI
- Runs
- Locally or Remotely
- Version
- 1.2.3
- Listing updated
Similar MCP servers
Install
Run in a terminal
claude mcp add kanopy --env KANOPY_TOKEN='<KANOPY_TOKEN>' -- docker run -i --rm -e KANOPY_TOKEN ghcr.io/acidvegas/kanopy:1.2.3Add to claude_desktop_config.json
{
"mcpServers": {
"kanopy": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"KANOPY_TOKEN",
"ghcr.io/acidvegas/kanopy:1.2.3"
],
"env": {
"KANOPY_TOKEN": "<KANOPY_TOKEN>"
}
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"kanopy": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"KANOPY_TOKEN",
"ghcr.io/acidvegas/kanopy:1.2.3"
],
"env": {
"KANOPY_TOKEN": "<KANOPY_TOKEN>"
}
}
}
}Add to .vscode/mcp.json
{
"servers": {
"kanopy": {
"type": "stdio",
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"KANOPY_TOKEN",
"ghcr.io/acidvegas/kanopy:1.2.3"
],
"env": {
"KANOPY_TOKEN": "<KANOPY_TOKEN>"
}
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"kanopy": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"KANOPY_TOKEN",
"ghcr.io/acidvegas/kanopy:1.2.3"
],
"env": {
"KANOPY_TOKEN": "<KANOPY_TOKEN>"
}
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"kanopy": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"KANOPY_TOKEN",
"ghcr.io/acidvegas/kanopy:1.2.3"
],
"env": {
"KANOPY_TOKEN": "<KANOPY_TOKEN>"
}
}
}
}Using several servers? Build one config file with the MCP config generator.
Configuration
KANOPY_TOKENAPI token from kanopy's Settings, API tokensRequiredKANOPY_AGENTAgent name shown in comments and history
About kanopy
Kanopy is a kanban board you run on your own machine, with a SQLite file by default or PostgreSQL for larger boards. Agents connect over MCP and can file tasks, claim work, move cards between To-Do, Doing and Done, and leave progress notes, while every change is attributed to both the agent and the person whose token it used.
The board is built around avoiding collisions when several agents work alongside people. Moving a task to Doing claims it for that agent session, so another session is told who holds it, and descriptions cannot be overwritten by an agent that has not read them first. A Claude Code session hook can also brief a new session with the open tasks for the current project folder.
It suits developers and small teams who already use coding agents and want a shared, self-hosted place for projects, chores and ideas, with push notifications through ntfy and a mobile-friendly interface.
Tools this server provides (10)
Taken from the project documentation; the running server may differ.
- list_projects
- Project tree with open task counts
- list_tasks
- Tasks filtered by project, team, labels, text, status, priority or due date; lists ready and recently changed tasks
- get_task
- Full tasks with description, links and recent comments, or an attachment's text
- save_task
- Create or update task fields, edit or append to the description, tick checklist items, comment
- save_project
- Rename or move, describe, set the local folder, archive
- comment
- Add a progress note to a task
- link
- Relate two tasks, or remove the relation
- labels
- List, create global type labels (admin), or delete labels
- inbox
- List unread notifications
- delete
- Move a task to the trash or delete an empty project; restore a deleted task
What it does
- Agent attribution on changes
- Comments and history entries record the agent name alongside the account whose token it used.
- Task claiming
- Moving a task to Doing claims it for one agent session so others do not duplicate the work.
- Session briefings
- A Claude Code hook hands a new session the open tasks for the linked project folder.
- Compact tool responses
- Tools return one line per task in plain text to keep agent context windows small.
- Scoped API tokens
- Tokens can be limited to certain projects and to a maximum role such as read-only or editor.
- Recurring tasks and history
- Tasks can repeat on a schedule and keep a history of changes, with recent description versions restorable.
- Self-hosted storage
- Runs as a single Python process with daily SQLite backups or an optional PostgreSQL database.
Good for
- Letting a coding agent file, update and complete tasks on a project board
- Preventing two agent sessions from working the same task at once
- Keeping a shared self-hosted board for a small team of people and agents
- Giving a new coding session the current open tasks for its project folder
- Tracking personal projects, chores and ideas on a phone-friendly board
Pricing
Open source
Prices change. Check acidvegas.github.io before you buy.
kanopy FAQ
How do agents connect to kanopy?
Agents connect to the MCP endpoint at /mcp over streamable HTTP, authenticated with a personal API token created in Settings. A stdio mode is also available for clients that launch local commands.
Where is the data stored?
Everything kanopy stores lives in one data folder on the host machine. SQLite boards are backed up daily, and PostgreSQL is available for larger teams.
What happens if two agents try to work the same task?
Moving a task to Doing claims it for that agent session. Another session trying to start it is told who already has it and only takes over if told to.
Can I give an agent read-only access?
Yes. API tokens can be set to a highest role of read-only, editor, manager or full access, and can be restricted to certain projects.
Is kanopy free?
Yes. kanopy is open source, so you can use and self-host it for free.
How do I add kanopy to Claude Code?
Run this command in your terminal: claude mcp add kanopy --env KANOPY_TOKEN='<KANOPY_TOKEN>' -- docker run -i --rm -e KANOPY_TOKEN ghcr.io/acidvegas/kanopy:1.2.3
What are the best alternatives to kanopy?
The closest alternatives to kanopy are Kanboard MCP, pkanban, Kanban AI, KanbanMCP and Runlane.
Where this comes from
- Description: written from the project's README, the project's website and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.