Plane MCP Server

What is Plane MCP Server?

A Model Context Protocol (MCP) server that provides tools and resources for interacting with Plane through AI agents. It supports multiple transports and can be used both locally and as a remote service.

How to use Plane MCP Server?

Use uvx (recommended, requires Python 3.10+) for local stdio transport, or npx with mcp-remote@latest (requires Node.js 22+) for remote HTTP/SSE transports. Configure authentication via environment variables for stdio, or use OAuth/PAT tokens for remote connections. Detailed JSON configuration examples for MCP clients are provided in the README.

Key features of Plane MCP Server

• Integrates with Plane APIs and services
• Supports stdio, SSE, and streamable HTTP transports
• Works locally and as a remote service
• Extensible; easy to add new tools and resources
• Uses Pydantic models for type safety and validation

Use cases of Plane MCP Server

• Manage projects, work items, cycles, and modules from an AI agent
• Automate creation and updates of intake work items and milestones
• Search and filter work items across a workspace
• Handle work item comments, links, relations, and properties programmatically
• Manage labels, states, work item types, and initiative planning

FAQ from Plane MCP Server

What authentication methods are supported?

The server supports API key + workspace slug (stdio), OAuth (remote HTTP/SSE), and Personal Access Token (remote HTTP with PAT headers).

What runtime requirements are needed?

Python 3.10+ for stdio transport (via uvx); Node.js 22+ for remote transports (via npx).

Where is the remote server hosted?

The hosted Plane MCP server is available at https://mcp.plane.so/ with separate endpoints for HTTP OAuth, HTTP PAT, and legacy SSE transport.

Are there any known limits?

For remote HTTP transport with PAT, the server validates redirect URIs against an allowlist. The SSE transport is legacy and deprecated in favor of HTTP transport.

How does logging work?

The server emits structured JSON logs per tool call, including tool name, duration, status, opaque user id, and workspace slug. Logging user PII requires explicitly setting LOG_USER_INFO="true".