Karea MCP

What starecz/karea-mcp MCP server does

The starecz/karea-mcp MCP server connects an MCP client to Karea, a task manager. Its README lists 49 tools for working with tasks, projects, categories, questions, resources, sessions, requisites, Jira links, and recaps. The supplied tool catalogue includes operations for creating, editing, closing, deleting, viewing, and listing work items.

Tasks can be identified by visual ID, name, or UUID in several operations. A task can include a status, priority, deadline, category, assignee, description, tags, notes, requisites, links, long-form markdown, and AI-oriented context. The server also supports quick logging for completed work and a separate operation for recording work currently in progress.

How it works

An MCP-compatible client starts the server locally with npm and passes Karea connection settings through environment variables. The client then discovers the available tools and can invoke them during a coding session.

Task listing defaults to open work and limits responses to 200 items. Status, category, priority, assignee, and search filters can narrow results on the server. Closed work can be queried with a status and an optional relative time such as 14d, 7d, or 24h; all requests all statuses. Batch task inspection is available through karea_view_tasks, which accepts up to 50 identifiers in one call.

The task knowledge model has separate storage areas. Markdown is intended for long-form research, requirements, designs, debugging records, and implementation notes. Context stores titled, AI-facing history about decisions, attempts, findings, and gotchas. Notes are intended for human-facing updates. Existing markdown should be read before replacing it, because the write operation overwrites the current document.

Setup and configuration

Install and run the starecz/karea-mcp MCP server with:

npx -y karea-mcp

For Claude Code, add an MCP entry to ~/.claude.json or use claude mcp add. The documented configuration sets the command to npx, passes -y and karea-mcp, and provides KAREA_API_KEY plus KAREA_URL. The example URL is https://karea.app.

For Cursor, add the server to .cursor/mcp.json in a project or to ~/.cursor/mcp.json globally. The documented Cursor example provides KAREA_API_KEY; it does not include KAREA_URL. The README directs users to the Karea dashboard API-key page to obtain a key.

Tools and capabilities

The starecz/karea-mcp MCP server supports these workflows:

• Create projects, categories, tasks, subtasks, questions, notes, and text resources.
• Edit or permanently delete tasks, projects, categories, questions, notes, and resources. Destructive operations require confirmation where documented.
• Mark individual or multiple tasks done, start an in-progress task, or log already completed work.
• Read and write task markdown, AI context, notes, requisites, and attached resources.
• Add, answer, edit, list, or remove open questions linked to projects or tasks.
• Link and unlink resources, Jira items, and sessions.
• Share projects with another user using owner, editor, commenter, or viewer roles.
• Produce recent activity recaps for standups or reviews.

The karea_ask tool sends a natural-language request to Karea’s AI assistant. According to the tool description, that assistant may read or modify tasks and consumes the account’s monthly AI usage allowance.

Limitations and notes

The server requires a Karea API key. The README does not describe pricing, API limits, or whether the Karea account or AI allowance is free or paid. The tool descriptions identify irreversible deletion for tasks, projects, categories, questions, notes, and resources; confirmation is required for the destructive operations where specified.

The provided material does not document a hosted remote endpoint. The documented installation runs the server locally through npx. The README names Claude Code and Cursor as configuration targets, while other MCP-compatible clients may work but are not individually configured in the supplied material.