Linear is a project management platform built for software teams. The linear mcp integration lets AI agents read issues, update tickets, create projects, and post comments directly inside your Linear workspace. Instead of copying ticket IDs into chat or switching between browser tabs, your agent operates inside the same system your team already uses for sprint planning, bug tracking, and roadmap management.
Linear's official MCP server is a hosted remote service. It follows the authenticated remote MCP specification, which means Linear manages the server centrally at mcp.linear.app. You do not install or host anything yourself. You authenticate, configure your client, and the agent gains access to your workspace data through standard OAuth flows.
How to Connect the Official Linear MCP Server
The setup path depends on your client. Claude and Cursor support native connectors. Other clients use the mcp-remote wrapper module.
Claude Desktop: Open Settings, navigate to Connectors, and add Linear. The OAuth flow opens in your browser. Log in with your Linear credentials and authorize the connection. Enterprise teams using Okta can enable MCP enterprise managed authentication in Linear settings and supply their Okta issuer URI, which lets supported clients authenticate users automatically through existing access policies.
Claude Code: Run this command:
claude mcp add --transport http linear-server https://mcp.linear.app/mcp
Then run /mcp inside a Claude Code session to complete OAuth authentication. After the handshake, your agent can list issues, create tickets, and update project status without leaving the terminal.
Cursor: Search for Linear in Cursor's MCP tools page, or click the official install link in Linear's documentation. Cursor handles the OAuth flow internally.
VS Code, Windsurf, and Zed: These clients need the mcp-remote module because they do not yet support authenticated remote MCP natively. Add this configuration:
{
"mcpServers": {
"linear": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.linear.app/mcp"]
}
}
}
For read-only access, replace the URL with https://mcp.linear.app/mcp/readonly. This endpoint exposes only read tools, which is useful if you want AI visibility into your workspace without write permissions. Alternatively, request only the read OAuth scope on the standard endpoint. Both approaches prevent the agent from creating, updating, or deleting any Linear data.
Compatible MCP Clients
| Client | Compatible | Auth Method |
|---|---|---|
| Claude Desktop | Yes | Native OAuth connector |
| Claude Code | Yes | OAuth via claude mcp add |
| Cursor | Yes | Native MCP directory install |
| VS Code | Yes | mcp-remote wrapper |
| Windsurf | Yes | mcp-remote wrapper |
| Zed | Yes | mcp-remote wrapper |
| Codex | Yes | Native codex mcp add |
| ChatGPT | No | Not supported |
What Your Agent Can Do in Linear
The official Linear MCP server exposes tools for the core objects in your workspace. Your agent can list and search issues, read projects and cycles, view comments and labels, and inspect team membership. On the write side, it can create issues, update status and priority, add comments, create projects, and modify issue relationships.
Linear publishes example prompt templates for common workflows. These are worth copying into your client:
Roadmap planning: Turn a planning document into a structured Linear project with milestones, issues, and relationships. The agent reads the document, proposes a project structure with clear titles and descriptions, and creates the issues after you approve. It organizes work into milestones when the document implies phases or deliverables.
Standup note automation: Paste raw standup notes and the agent matches them to existing Linear issues by ID, title, or owner, then posts concise progress comments on each ticket. Ambiguous notes are flagged rather than guessed, so your issue history stays clean.
Bug investigation: Starting from a Linear issue, the agent searches your codebase for likely root causes and posts a summary of findings back to the ticket. It flags uncertainty when evidence is weak rather than inventing explanations.
Cycle summarization: The agent gathers all issues completed in a recent cycle and produces a summary of themes, blockers, and outcomes for team review. The output is factual and avoids overstating progress on work that is still in flight.
Timeline generation: The agent builds a chronological history of work around a specific topic, showing when issues were created, when direction changed, and when major decisions were recorded. Gaps are called out rather than inferred.
Quality and Security Notes
Because the official Linear MCP server is a hosted service managed by Linear, it does not appear in the MCPVault GitHub index. MCPVault catalogs open-source repositories with installable code, and this is a centrally managed endpoint without a public repository.
If you need a community alternative that is indexed in the vault, linear-mcp-server by jerhadf is available. It carries a Grade C on MCPVault with a status of stale, has 347 stars, and supports Claude Desktop. It is a local stdio server rather than a hosted remote endpoint, which means you host it yourself and manage the Linear API key directly.
See Linear MCP Server on MCPVault for grades, tools and install details.
For the official server, the security model is straightforward. OAuth tokens are per-workspace. If you need to connect multiple Linear workspaces, you must maintain separate authentication contexts because reconnecting within the same session does not switch workspaces. Use the MCP_REMOTE_CONFIG_DIR environment variable to isolate configs:
MCP_REMOTE_CONFIG_DIR=~/.mcp-auth/workspace-a npx mcp-remote https://mcp.linear.app/mcp
Read-only endpoints and scoped OAuth tokens let you grant AI access at the right level of privilege. For teams with strict compliance requirements, the read-only endpoint is the safer starting point. You can always upgrade to read-write later by reconnecting with the full scope.
Frequently Asked Questions
Can I use the Linear MCP server with multiple workspaces?
Yes, but each workspace needs its own authentication context. The Linear MCP server authenticates via OAuth, and reconnecting within the same session does not switch workspaces. For clients using mcp-remote, set a different MCP_REMOTE_CONFIG_DIR for each workspace and authenticate them separately. Clients that manage their own auth storage, like Claude Desktop, may have workspace-switching options in their own documentation.
What is the difference between the official Linear MCP and community alternatives?
The official server is hosted by Linear at mcp.linear.app and follows the authenticated remote MCP spec. It receives updates automatically and supports native OAuth in Claude and Cursor. Community alternatives like linear-mcp-server run locally through stdio, require self-hosting, and may lag behind Linear API changes. The official server is the recommended choice for production use. See how MCPVault grades servers for the methodology behind quality scores.
Does the Linear MCP server support read-only access?
Yes. Connect to https://mcp.linear.app/mcp/readonly for an endpoint that only exposes read tools. Alternatively, use the standard /mcp endpoint but request only the read OAuth scope. Both approaches prevent the agent from creating, updating, or deleting any Linear data.
Can I authenticate with an API key instead of OAuth?
Yes. The Linear MCP server supports API key authentication as an alternative to OAuth. Generate an API key from Linear Settings under Account, then pass it as a bearer token. This is useful for automation scenarios or clients that do not support OAuth flows. You can also use API keys with the mcp-remote module by setting the appropriate header.
Your server in the vault.
If you built an MCP server, claim the listing and get it in front of developers who are actively looking. Free to claim. Verification is free during early access and earns a do-follow link to your project.
Claim your listing or submit a server if it is not indexed yet.