VibeKanban MCP Server

VibeKanban: An AI‑Powered Kanban Board for Modern Development Workflows

VibeKanban is designed to give AI assistants, such as Claude or Cursor’s MCP agents, a tangible, structured view of your project’s progress. By exposing a Kanban board through the Model Context Protocol (MCP), it solves the common pain point of opaque agent activity—developers often struggle to understand what an automated agent is doing when it “fixes a bug” or “builds a feature.” VibeKanban bridges this gap by providing a visual, stateful workspace that the agent can read from and write to just like any human teammate.

At its core, VibeKanban offers a lightweight Flask backend paired with a responsive HTML/JavaScript front end built on Tailwind CSS. The server manages projects, tickets (bugs, stories, tasks, spikes), and priorities, exposing these entities via MCP commands such as , , or . This command set enables agents to query the current board state, spawn new work items, and advance tickets through stages—all while keeping the UI in sync for human observers. The MCP integration is deliberately simple: a single command line launches the server, and Cursor or any other MCP‑compatible client can register it as a global server.

Key capabilities include drag‑and‑drop Kanban visualization, real‑time metrics (DORA indicators like deployment frequency and mean time to recovery), and a clean API that lets agents act as if they were human developers. Because the board is stateful and persists in a local SQLite database, agents can pick up where they left off even after restarts. The server’s metrics layer also allows developers to audit agent performance, correlating automated actions with measurable outcomes.

Typical use cases span from continuous integration pipelines that auto‑create tickets when tests fail, to pair programming scenarios where an agent proposes code changes and the human reviewer can immediately see the work item on the board. In distributed teams, VibeKanban ensures that every stakeholder—human or AI—has a single source of truth for task status, reducing miscommunication and accelerating delivery. Its lightweight footprint means it can run locally or in a containerized environment, making it suitable for both solo developers and larger organizations.

In summary, VibeKanban transforms abstract AI actions into concrete board updates, providing transparency, traceability, and actionable metrics. For developers building or integrating MCP agents, it offers a ready‑made interface that turns code generation and bug fixing into observable, trackable work items—exactly the kind of context that makes AI assistants feel like true teammates.