MCP Kanban Memory

MCP Kanban Memory is an MCP server that provides a kanban-based task management and state retention system for AI-driven workflows. It connects large language model clients to an embedded SQLite database, giving agents a persistent, visual structure to manage multi-step engineering and research projects. Software developers and technical users utilize this server to overcome context window limitations by offloading plans and task states across separate chat sessions. MCP Kanban Memory enforces core kanban principles such as column capacity and work-in-progress (WIP) limits, preventing agents from taking on too many tasks concurrently. The system comes paired with an optional standalone web user interface running locally, allowing human supervisors to inspect the current board state, view task details in markdown, and manually move or update cards. Through dedicated tools and predefined prompts, the server enables language models to break down high-level project goals into discrete tasks, register progress, move work between columns, and cleanly resume prior projects on demand.

Category: AI Memory & Context

Tags: kanban, productivity, task-management, workflow

Visit MCP Kanban Memory

How to install and configure MCP Kanban Memory

1. Clone the repository from https://github.com/eyalzh/kanban-mcp. 2. Install dependencies and build the database and server packages: sh npm ci --prefix shared/db npm ci --prefix mcp-server npm run build --prefix shared/db npm run build --prefix mcp-server 3. Add the server configuration to your MCP client configuration file (such as Claude Desktop): json { "mcpServers": { "kanban-mcp": { "command": "node", "args": [ "/path/to/repo/mcp-server/dist/server.js" ], "env": { "MCP_KANBAN_DB_FOLDER_PATH": "/path/to/db" } } } } 4. Replace /path/to/repo with your local repository path and /path/to/db with a folder where the SQLite database files can be created.

What you can do with MCP Kanban Memory

• Creating structured project roadmaps by prompting the model to break high-level requirements into discrete kanban tasks across defined columns. - Tracking task execution across disconnected sessions by having the AI query existing boards and pick up the next available item. - Enforcing work-in-progress limits during autonomous coding workflows to prevent the assistant from juggling too many concurrent modifications. - Viewing real-time task progress through the local web dashboard while allowing human operators to manually update or reorder cards.

Key facts

• https://github.com/eyalzh/kanban-mcp
• AI Memory & Context, Project & Task Management
• kanban, productivity, task-management, workflow

Part of MCP Servers

How do I install MCP Kanban Memory?

Clone the GitHub repository, install the dependencies, and build the shared database and mcp-server directories using npm. Afterwards, add the kanban-mcp entry to your client configuration, pointing node to the compiled server.js file and setting the MCP_KANBAN_DB_FOLDER_PATH environment variable to a writable directory.

What can MCP Kanban Memory do?

It provides tools for creating kanban boards, adding markdown-formatted tasks, moving cards between columns with WIP limit enforcement, deleting tasks, and inspecting board states. It also includes prompts to initialize projects and resume ongoing work, backed by SQLite storage and an optional browser UI.

Which MCP clients work with MCP Kanban Memory?

It works with any MCP-compliant client that supports local stdio command execution and environment variable configuration, including Claude Desktop. You can also run it inside a Docker container by configuring the docker command in your MCP client settings.

How do I access the web UI for MCP Kanban Memory?

Build the web-ui and web-server packages using npm ci and npm run build. Then launch the web-server with the MCP_KANBAN_DB_FOLDER_PATH environment variable set to your database directory. The dashboard is accessible locally at http://localhost:8221 in any browser.