TaskBoardAI MCP Server

Overview

TaskBoardAI is a lightweight, file‑based Kanban board that serves as an AI‑friendly project management hub. By storing boards in plain JSON, it gives AI agents full visibility into the current state of a project without needing to parse proprietary APIs or databases. The MCP server exposes CRUD operations on these JSON files, allowing assistants like Claude to create, read, update, or delete boards and cards directly from the conversation. This eliminates the need for manual file manipulation and keeps the board in sync with the assistant’s evolving context.

The server also powers a user‑facing web interface that supports drag‑and‑drop sorting, card and column manipulation, and a dropdown for quick board selection. This Human‑In‑The‑Loop (HIL) interface lets developers and stakeholders keep the board up to date while an AI agent iterates on tasks in the background. The combination of a web UI and an MCP API makes TaskBoardAI a bridge between human workflows and AI automation.

Key capabilities include:

• Markdown‑rich cards for detailed task descriptions, code snippets, and links.
• Subtasks that can be individually marked complete, enabling fine‑grained progress tracking.
• Tags and dependencies to organize related work and enforce ordering constraints between cards.
• Next‑step tracking at the board level to surface upcoming priorities for quick reference.
• Webhooks that can notify external services when cards change, allowing integration with CI/CD pipelines or issue trackers.
• MCP‑based AI integration that lets an assistant manipulate the board through simple JSON requests, keeping project context in sync with AI reasoning.

Real‑world scenarios where TaskBoardAI shines include:

• Agile development teams that need an AI to propose sprint backlogs or re‑prioritize tasks based on new requirements.
• Documentation workflows where an assistant drafts or updates markdown content directly in card bodies.
• Continuous delivery pipelines that trigger webhook notifications to update deployment boards when builds succeed or fail.
• Remote collaboration where stakeholders can view and edit the board in real time while an AI agent handles routine updates or status summaries.

Integrating TaskBoardAI into an AI workflow is straightforward: a developer can start the MCP server, point their assistant to the board’s JSON endpoint, and issue commands such as “Add a new card titled ‘Implement OAuth’ to the Backlog column.” The assistant receives the updated board in its context, can generate further suggestions, and then write back changes through the MCP API. This tight coupling of human oversight (via the web UI) and AI automation (via MCP) provides a powerful, low‑friction loop for managing complex projects.