Taskboard is an MCP server tool by Tcarac, designed for AI-native project management on the mcp platform. Built with Go, it distributes as a single binary, installable via Homebrew or source. It includes an embedded frontend, allowing individual developers and small teams to manage projects and tickets with AI assistants. This local, self-hosted solution ensures data remains on your machine, connecting AI agents to project tasks.

What does Taskboard connect to?

Taskboard connects to a local SQLite database for all its data storage needs, ensuring project information remains securely on the user's machine. It functions as a Model Context Protocol (MCP) server, specifically tailored for AI assistants in project management. This architecture allows AI clients, such as opencode and claude, to interact directly with project data stored locally. Its design as a local, self-hosted tool reinforces that your data stays private and on your system.

Core features and tools

Taskboard exposes 22 tools for AI-native project management via Model Context Protocol. While the specific names of these tools are not explicitly provided, it offers capabilities for AI assistants to manage projects, tickets, and teams directly. It enables AI agents to create tickets with subtasks by analyzing various project requirements, such as for API endpoints, email templates, or UI screens. AI agents can also analyze the active task board to identify blockers and break down "in progress" tasks into concrete next steps.

Who is it for?

Taskboard targets individual developers or small teams that actively utilize AI assistants for project management tasks. It empowers AI agents to handle project tasks, ticket creation, and team coordination within a local development environment. The tool supports scenarios where AI analyzes project requirements to generate detailed subtasks or reviews project progress to identify concrete next steps. Skip this if your project requires a distinct, externally managed MCP-level authentication mechanism.

How to get started

No credentials are required for its MCP server, relying instead on local system access for security. The server supports JSON-RPC over stdio as its transport protocol for AI clients.

Use cases

• AI assistants manage projects, tickets, and teams directly.
• AI agents create tickets with subtasks by analyzing project requirements (e.g., for API endpoints, email templates, UI screens).
• AI agents analyze the task board to identify blockers and break down "in progress" tasks into concrete next steps.

Taskboard offers a self-hosted Model Context Protocol server for AI-driven project management, keeping all data locally. Its Go-based single binary architecture and Homebrew installation make it accessible for developers. It supports AI clients for direct project and ticket handling, and agents can analyze tasks to ensure progress. Try having an AI agent create tickets with subtasks by analyzing project requirements.