mcp-linear

mcp-linear

MCP.Pizza Chef: tacticlaunch

One hundred and ninety-five separate actions ship here, far more than most: reading and writing issues, comments, projects, cycles, milestones, roadmaps, documents, saved views, customer records, and notifications. Everything is switched on all the time, including forty-one actions that archive, delete, or remove things, and there is no look-but-don't-touch mode, so the assistant can retire a project or delete a comment on its own reading of your request. Sign in with a personal Linear key or the built-in login step.

Coding
Communication

Use This MCP server To

File a bug in the right team without leaving my chat Move a ticket to In Progress and assign it to someone Summarize what my team shipped this cycle Draft this week's project update from what Linear already knows Find every open issue blocking a milestone Leave a comment on a ticket asking for more detail

README

Linear App Icon

MCP Linear

A Model Context Protocol (MCP) server for the Linear GraphQL API, built for real project-management workflows — not just basic issue CRUD.

MCP Linear npm version

Features

MCP Linear bridges AI assistants and Linear by implementing the MCP protocol. With it you can:

  • Retrieve issues, projects, teams, cycles, milestones, roadmaps, customers, customer needs, and workspace/project/initiative/team/issue/release/cycle documents
  • Create and update issues, change status, assign, and comment
  • Manage projects, full diff-aware project and initiative update lifecycles, milestones, roadmaps, saved views, and favorites
  • Create and manage workspace webhooks, including updates and signing-secret rotation
  • Prepare OAuth app manifests and authorization URLs, issue scoped client-credentials tokens, or manage child OAuth apps when authenticated as a managing OAuth application
  • Work with templates, custom fields, and attachments
  • Work with customer records, customer statuses/tiers, and customer needs linked to issues or projects
  • Read notifications, subscriptions, sessions, audits, and integrations without leaving MCP
  • Inspect rate-limit and server health before running heavy planning sessions

See TOOLS.md for the full inventory.

MCP-native resources and prompts

The server exposes MCP resources and prompts in addition to tools, including:

  • Resources: linear://viewer, linear://organization, linear://teams, linear://projects, linear://project/{id}, linear://project/{id}/issues, linear://project/{id}/documents, linear://issue/{id}, linear://document/{id}, linear://roadmap/{id}, linear://milestone/{id}, linear://rate-limit
  • Prompts: summarize-project-status, draft-project-update, triage-issue, summarize-document

Example prompts

Once connected, you can use prompts like:

  • "Show me all my Linear issues"
  • "Create a new issue titled 'Fix login bug' in the Frontend team"
  • "Change the status of issue FE-123 to 'In Progress'"
  • "Assign issue BE-456 to John Smith"
  • "Show all open issues in this project grouped by milestone and cycle"
  • "Draft a weekly project update from the current Linear state"
  • "Find the newest documents related to a project and summarize the key decisions"
  • "Show the pinned documents and links on this team's home page"
  • "Create a document for ENG-123 with resource ordering metadata"
  • "Get the latest project update diff and archive an outdated update"
  • "Show customer needs for this project and mark the important ones"
  • "Create an initiative update and hide the generated diff from the update body"
  • "Prepare a private OAuth app for my GitHub issue pipeline with client credentials enabled"
  • "Issue a narrowly scoped client-credentials token for that GitHub pipeline"
  • "Create a webhook for Issue and Comment events, then rotate its signing secret"

Installation

Authentication

Personal API key (default)
  1. Log in to your Linear account at linear.app
  2. Click on your organization avatar (top-left corner)
  3. Select Settings
  4. Navigate to Security & access in the left sidebar
  5. Under Personal API Keys click New API Key
  6. Give your key a name (e.g., MCP Linear Integration)
  7. Copy the generated API token and store it securely — you won't be able to see it again

Personal API keys support the normal Linear and workspace-webhook tools. They cannot call Linear's alpha managed-child-OAuth-application API because that API requires the caller itself to be an OAuth application. With a personal API key, linear_generateOAuthApplicationSetup still prepares an official manifest and pre-filled Linear setup URL for an administrator to confirm.

OAuth access token (managed OAuth applications)

To let MCP actually create and manage child OAuth applications, authenticate it with an access token belonging to a Linear OAuth application that is eligible to manage those child applications:

export LINEAR_OAUTH_ACCESS_TOKEN=YOUR_OAUTH_ACCESS_TOKEN
mcp-linear

Or pass --oauth-token YOUR_OAUTH_ACCESS_TOKEN. Explicit command-line credentials take precedence over environment variables; when both environment credential types are present, OAuth authentication is selected. See Linear's OAuth documentation and OAuth application manifests.

Each MCP server process uses one Linear credential. If the managing app token uses actor=app (which cannot receive admin) and you also need admin-scoped workspace-webhook tools, configure two MCP server entries: one with the managing OAuth token for child-app operations and one with a workspace administrator's personal API key for workspace webhooks. A user-actor OAuth token carrying admin can cover the webhook side instead.

OAuth scopes are selected when an authorization URL or client-credentials token is requested; they are not mutable fields on an OAuth application. The MCP validates current Linear scopes, prepares authorization URLs, and can issue app-actor tokens with linear_createOAuthClientCredentialsToken. For GitHub-hosted pipelines, enable the client_credentials grant and request the narrowest useful scope, such as issues:create.

Client-credentials tokens normally expire after 30 days and have no refresh token. Linear permits multiple active tokens only while they use the same scope set; requesting a different scope set revokes the application's existing app-actor tokens. The token tool therefore requires both confirmSecretExposure: true and confirmScopeChangeRisk: true.

Creating an OAuth app and rotating OAuth or webhook secrets returns one-time secret material through MCP. Those tools require confirmSecretExposure: true; move returned values directly into a secret manager such as GitHub Actions secrets and do not paste them into source control or logs.

Webhook URLs must be publicly reachable HTTPS endpoints. Validation rejects credentials in URLs and obvious loopback, private-network, link-local, and local-hostname destinations.

Installing via add-mcp (Recommended)

add-mcp installs the server into Claude Code, Cursor, Codex, VS Code, Claude Desktop, and many other MCP-aware agents with a single command:

npx add-mcp @tacticlaunch/mcp-linear --env LINEAR_API_TOKEN=YOUR_LINEAR_API_TOKEN

Add -g to install globally instead of into the current project. See the add-mcp docs for the full agent list and flags.

Manual configuration

Add the following to your MCP settings file:

{
  "mcpServers": {
    "linear": {
      "command": "npx",
      "args": ["-y", "@tacticlaunch/mcp-linear"],
      "env": {
        "LINEAR_API_TOKEN": "<YOUR_TOKEN>"
      }
    }
  }
}
Client-specific configuration locations
  • Cursor: ~/.cursor/mcp.json
  • Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Claude VSCode Extension: ~/Library/Application Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json
  • GoMCP: ~/.config/gomcp/config.yaml

Manual run

Prerequisites:

  • Node.js (v20+)
  • NPM or Yarn
  • Linear personal API key or OAuth access token
# Install globally
npm install -g @tacticlaunch/mcp-linear

# Or clone and install locally
git clone https://github.com/tacticlaunch/mcp-linear.git
cd mcp-linear
npm install
npm link  # Makes the package available globally
Running the server

Run the server with your Linear API token:

mcp-linear --token YOUR_LINEAR_API_TOKEN

Or with a managing OAuth application's access token:

mcp-linear --oauth-token YOUR_OAUTH_ACCESS_TOKEN

Or set the token in your environment and run without arguments:

export LINEAR_API_TOKEN=YOUR_LINEAR_API_TOKEN
mcp-linear

Validation

The default validation path is:

npm test
npm run build

npm test runs Jest unit tests and an official MCP SDK smoke test against the built stdio server, covering tool, resource, and prompt registration plus host-compatible schema emission.

Development

See DEVELOPMENT.md for local development details.

Links

tacticlaunch/cursor-memory-bank — If you are a developer seeking to enhance your workflow with Cursor, consider giving it a try.

License

This project is licensed under the MIT License — see the LICENSE file for details.

mcp-linear FAQ

Can it delete or archive things by mistake?
Yes, and this is the main thing to know. Forty-one of the one hundred and ninety-five actions archive, delete, or remove something, they are always available, nothing asks you to confirm, and there is no setting that limits it to reading. Connect it with an account whose access matches how much you trust it.
What credential does it need, and does it cost anything?
Either a personal Linear key you create under Settings, then Security and access, then Personal API Keys, or the sign-in step built into the tool. Both are free with your existing Linear account, and the connection inherits exactly the permissions that account already has.
Can I use this to file a bug without opening Linear?
Yes. Describe the problem and name the team, and it creates the issue, sets the status, assigns it, and adds a comment if you ask.
Which apps does it work in?
Claude Desktop, Cursor, and VS Code have documented setups, and it installs with a single command. Any other app that accepts a pasted server entry will work.
Does offering nearly two hundred actions cause problems?
It can. Some apps slow down, clutter up, or hit their own limits when a single connection offers this many choices, and there is no way to switch off the ones you do not want.
Can it really create and manage sign-in applications, as the description suggests?
Only partly, and the project says so itself. A personal key cannot create or manage those applications; the relevant action just prepares a setup link for an administrator to confirm. Full management needs a token belonging to a managing application.
How hard is setup?
Easy. One command installs it, then you paste a short block with your key into your app's settings, or run the sign-in step instead.
Is it actively maintained?
Yes. Both the code and the published package were updated in July 2026, at version 1.4.1, and it carries a permissive MIT licence.