taskboard, from Tcarac, is a self-hosted project manager that connects Kanban boards to MCP for AI access. It offers a web-based drag-and-drop board, a command-line interface, and programmatic agent access so assistants can list, create, and update tasks. Local SQLite storage and single-binary distribution support zero-configuration launches and local data custody. The tool targets developers and technical teams who need local control and AI-assisted task automation.

Where the tool sits in a developer workflow

The tool combines visual task tracking with terminal-driven control, providing an interactive Kanban board with Todo, In Progress, and Done columns plus multi-board support. The interface supports drag-and-drop organization and the CLI exposes full task and project management. Use cases include sprint tracking, backlog grooming, and coordinating work when developers prefer terminal commands alongside a web UI. A short list of visible elements:

• Drag-and-drop columns for status
• Multiple boards per project
• Full CLI task control

How AI integration behaves and its limits

The tool runs an MCP server so MCP-compliant agents can programmatically read, create, and modify tasks; it explicitly cites compatibility with clients such as Claude Desktop. AI-driven features require an external agent connection and therefore network access, while the core board and CLI work offline. Agent edits are automated operations, so outputs from agents should be reviewed before accepting substantive project changes.

Deployment, privacy posture, and target suitability

Distributed as a cross-platform single binary, the app uses an embedded SQLite database to store all data locally in a bundled file, which simplifies setup for technical operators. That local-first model keeps project data on the machine unless a user explicitly shares it with an AI provider. The tool suits developers and power users; non-technical users may encounter a learning curve around MCP client setup and agent workflows.

A developer-focused choice for local, agent-aware task management

The tool is a practical option for developers and technical teams who require local custody and programmatic agent access while accepting a developer-oriented workflow and setup. Expect to pair agent-driven updates with human review, since agent changes automate routine task edits but still need oversight for complex decisions. For teams prioritizing on-device control and scripted integrations, the app fits those operational needs well.