Linear MCP Server

Overview

Linear MCP Server is the official server published by Linear that exposes Linear's project and issue tracking APIs as structured MCP tools, giving agents direct read and write access to issues, projects, cycles, teams, and comments inside a Linear workspace. Agents can create, update, triage, and close issues; manage project and cycle state; post and retrieve comments; and search across the workspace — all without leaving the agent session or writing custom API wrappers. The server ships with both stdio and HTTP transports, making it suitable for local Claude Desktop and Cursor setups as well as remote multi-machine deployments. It pairs naturally with the GitHub MCP server to close the full issue-to-PR loop on Linear-led engineering teams, and works best with Claude Desktop, Claude Code, and Cursor as the agent host.

Best for

Best for
Linear-first product and engineering teams that want agents to automate issue triage, project tracking, and cycle management across the full Linear workspace

Use cases

• Triage and assign incoming Linear issues from an agent that reads, labels, and comments automatically
• Generate weekly cycle summaries and status reports from live Linear data
• Create and update issues directly from a coding agent when a bug or task is identified
• Bridge GitHub PR activity back into the correct Linear issue to keep project state in sync
• Search across projects and teams to surface relevant prior work before starting a new task

Notes

What it does

Linear MCP Server exposes Linear’s project and issue tracking APIs as a complete set of MCP-native tools. The core tool surface covers issues (create, update, assign, close, search, list by team or project), projects (create, update, list, retrieve milestones), cycles (list, retrieve active cycle, add issues to a cycle), teams (list, retrieve members and settings), and comments (create, list, retrieve on any issue). Agents get structured JSON responses they can reason over, filter, and act on without parsing raw REST responses.

How to deploy

Add the server to your MCP host config (e.g., claude_desktop_config.json under mcpServers) using the Linear-provided entry point, and set the LINEAR_API_KEY environment variable to a Linear personal API key generated from your workspace settings. For HTTP transport, run the server as a persistent process and point your MCP host at the HTTP endpoint. No additional dependencies or infrastructure are required for the stdio path.

Best practices

Pair Linear MCP with the GitHub MCP server in the same agent session to automate the full engineering workflow: the agent can move a Linear issue from triage to in-progress, open a GitHub branch, and mark the issue complete when the PR merges — all in one loop. Use a dedicated Linear API key scoped to the minimum required permissions, and consider creating a bot user in Linear for agent-driven writes so human and agent activity are clearly separated in the audit log.