Planka

What the planka MCP server does

The planka MCP server gives an MCP-compatible agent access to PLANKA kanban data and actions. It works with projects, boards, lists, cards, tasks, labels, and comments, allowing an agent to inspect existing work and make changes through structured tool calls.

Navigation tools return the project, board, and list hierarchy or retrieve a board with its cards, lists, and labels. Card operations cover creation, property updates, movement between lists and positions, detailed retrieval, and deletion. A new card can optionally include tasks at creation time.

Task tools manage checklist items on cards. Agents can add several tasks, change a task name or completion state, and remove tasks. Label tools handle board labels and the labels assigned to individual cards. Comment tools add comments and retrieve all comments associated with a card. List management supports creating, updating, and deleting lists.

How it works

The server runs as an MCP process and calls the PLANKA API using the configured agent account. Inputs are validated with Zod, and the implementation is intended for PLANKA 2.0 API behavior. That includes supplying the required card type during card creation, using the PLANKA 2.0 card-label endpoint, and handling optional fields without requiring them in every request.

An agent can first inspect the available workspace with planka_get_structure, locate a board or list, and then use the relevant mutation tool. For example, a workflow can create a card with checklist items, move the card as work progresses, apply labels, and append status comments. The server does not describe a separate synchronization layer or local data store; its documented role is to expose PLANKA operations over MCP.

Setup and configuration

Install the package with npm or invoke it directly with npx:

npm install @gogogadgetbytes/planka-mcp
npx @gogogadgetbytes/planka-mcp

Running the package requires three environment variables:

• PLANKA_BASE_URL: The URL of the PLANKA server.
• PLANKA_AGENT_EMAIL: The email address for the agent user.
• PLANKA_AGENT_PASSWORD: The agent user's password.

The README provides configuration examples for Claude Desktop and Claude Code. Both launch the package through npx and pass these values in the server's environment. The supplied credentials therefore need an agent account that can access the intended PLANKA workspace.

For local development, the repository instructions use npm install, npm run build, and npm test after cloning the project.

Tools and capabilities

The planka MCP server groups its tools by the type of PLANKA object they operate on:

• Navigation: inspect the project, board, and list hierarchy or fetch a complete board view.
• Cards: create, read, update, move, and delete cards; creation may include tasks.
• Tasks: add checklist items, update their names or completion status, and delete them.
• Labels: create, update, or delete board labels and add or remove labels on cards.
• Comments: add comments to cards and retrieve their comments.
• Lists: create, update, or delete lists.

These operations support common agent workflows such as finding a destination list, creating a work item with a checklist, moving it through a board, and recording progress in comments.

Limitations and notes

The documentation targets PLANKA 2.0 and specifically calls out differences from version 1.x. Compatibility with PLANKA 1.x is not stated. The server also requires PLANKA URL and agent credentials; without all three configuration variables, the documented setup is incomplete.

The README explicitly documents Claude Desktop and Claude Code configuration. It does not provide client-specific setup instructions for other MCP clients.