Linear is a Productivity MCP server that lets Claude Code, Cursor, Windsurf and any MCP-compatible AI agent issue tracking — projects, issues, cycles. Install in 1 minute with mcpizy install linear.
Connect your Linear account once — MCPizy stores the credentials encrypted and uses them whenever you run a recipe in managed mode.
mcpizy install linearnpx -y @linear/mcp-serverlist_issuesList issues with team, status, and assignee filters
Inputs
teamIdstringoptionalstatestringoptionalget_issueFetch the full details of an issue
Inputs
idstringrequiredsave_issueCreate or update an issue
Inputs
titlestringrequiredteamIdstringrequiredlist_projectsList projects in the workspace
list_teamsList all teams in the workspace
save_commentAdd a comment to an issue
Inputs
issueIdstringrequiredbodystringrequiredlist_cyclesList current and upcoming cycles for a team
Inputs
teamIdstringrequiredWorks identically across clients. Only the config file path differs.
~/.claude.json{
"mcpServers": {
"linear": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.linear.app/sse"
]
}
}
}.cursor/mcp.json{
"mcpServers": {
"linear": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.linear.app/sse"
]
}
}
}~/.codeium/windsurf/mcp_config.json{
"mcpServers": {
"linear": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.linear.app/sse"
]
}
}
}Install the Linear MCP and authenticate via OAuth (one-click flow)
Paste any of these prompts into Claude Code, Cursor or another MCP-compatible client.
“What issues are assigned to me in the current cycle?”
Uses: list_issues
“Create a bug ticket titled 'Login 500 error on Safari' in the Frontend team”
Uses: save_issue
“Summarise all In Progress issues in the Platform project”
Uses: list_projects, list_issues
“Comment 'Waiting on design' on issue ENG-421”
Uses: save_comment
Linear is the issue tracker that ships the cleanest MCP server for AI coding agents. The Linear MCP exposes the same primitives the Linear web app and CLI use — issues, projects, cycles, comments, labels — but as tool calls Claude Code, Cursor, Windsurf and Claude Desktop can chain together. The practical effect is that you can stop pasting issue links into chat and instead say "open the three highest-priority bugs in the Auth project, summarise the last comment on each, and draft a fix for the first one." The agent walks the graph itself.
Where Linear MCP earns its keep: tight feedback loops where the AI needs to read context (recent comments, linked PRs, cycle progress) before doing anything. We use it daily to triage backlog grooming — the agent reads the cycle, finds stale issues, asks for status, and updates labels — work that would otherwise be 30 minutes of clicking. The token cost is dominated by issue bodies; if your team writes long Linear descriptions, expect ~2-4k input tokens per issue inspected. Authentication is a single API key per user (no OAuth dance), which makes the server a clean fit for agent runners that don't have a browser.
Compared to writing your own Linear API wrapper, the MCP layer pays off because every MCP-compatible client (Claude Code, Cursor, Windsurf, VS Code Copilot Chat, custom agents on the SDK) gets the same tools for free. The trade-off is that the official MCP currently exposes ~7 high-level tools — enough for 95 % of read/update flows, but if you need GraphQL-level access (custom queries on a specific view, attachment uploads), you'll still drop down to the REST API.
The Linear MCP runs as a remote server in newer versions — make sure your client supports remote MCP transport (Claude Code 1.0.50+, Cursor 0.45+). Older versions only handle stdio servers and will silently skip Linear.
Workspace-level vs team-level scoping confuses people: the API key authorises against a workspace, but most useful queries (cycles, projects) need a team identifier. If your prompt produces "team not found" errors, ask the agent to call `list_teams` first and pin the team key in subsequent calls.
Rate limits are gentle for human use but can be hit by aggressive agent loops — a 10-issue triage that comments on each one will burn through ~50 API calls. If you see 429s, batch the updates with `update_many` style prompts rather than one-by-one.
Honest pros/cons against the closest productivity MCP servers.
| Server | Strengths | Trade-offs |
|---|---|---|
| GitHub MCP | Free, deeper code-side integration (PRs, files, search) | No native cycles/sprint concept — issue tracking is shallower |
| Jira MCP (community) | Mature for enterprises already on Atlassian | Slower API, OAuth flow harder to automate, fewer maintained MCP forks |
| Notion MCP | Works if your team uses Notion for planning instead of Linear | Page-tree model is awkward for issue workflows, no cycles primitive |
If Linear doesn't fit your stack, these Productivity MCP servers solve similar problems.
The Linear MCP server is an Productivity Model Context Protocol server that lets Claude Code, Cursor, Windsurf, VS Code with Copilot, and other MCP-compatible AI agents issue tracking — projects, issues, cycles. It exposes Linear's capabilities as tools the AI can call directly from your editor or CLI.
The fastest way is the MCPizy CLI: run `mcpizy install linear` and MCPizy will add the server to your `.claude.json` automatically. You can also install it manually by adding an entry under `mcpServers` in `.claude.json` with the command `npx -y @linear/mcp-server` and restarting Claude Code.
Yes. The Linear MCP server is free and open source (see the GitHub repository linked on this page). You may still need a Linear account or API key to connect the server to the underlying service, but the MCP layer itself has no MCPizy subscription cost.
Yes. Any MCP-compatible client works — including Claude Code, Claude Desktop, Cursor (via `.cursor/mcp.json`), Windsurf, VS Code with Copilot Chat, and custom agents built on the MCP SDK. The same install command targets all of them; only the config file path differs.
Once installed, your AI agent can issue tracking — projects, issues, cycles directly inside your conversation. Typical use cases include asking Claude Code or Cursor to run Linear operations, inspect results, chain Linear with other MCP servers (see our Workflow Recipes), and automate repetitive productivity tasks without leaving your editor.