Twist MCP Server

Overview

The Twist MCP Server bridges the gap between AI assistants and the collaborative workspace of Twist, a modern team communication platform. By exposing a rich set of MCP tools that mirror the native Twist REST API, this server allows Claude and other AI agents to perform real‑time workspace management tasks—such as retrieving inbox contents, creating or updating threads, and adjusting read states—without leaving the assistant’s conversational context. For developers building productivity workflows, this means an AI can act as a full‑fledged Twist bot: automatically triaging inboxes, moving conversations between channels, or flagging important messages on behalf of a user.

At its core, the server implements a straightforward authentication scheme: an OAuth‑2 token with full workspace scope is supplied via environment variables, giving the MCP server unrestricted access to all user‑level Twist resources. While this approach is currently intended for testing, future releases will replace it with a more secure OAuth flow. The server runs on Python 3.10+ and can be launched through the UV package manager, making it lightweight yet powerful enough for production use once fully tested.

Key capabilities include:

• Inbox management: Retrieve, archive, unarchive, and mark threads as read or unread. Bulk operations such as archiving all recent threads or clearing unread counts are also supported.
• Thread lifecycle: Create, update, delete, pin, star, and move threads across channels. The server exposes fine‑grained controls for read status, muting, and ownership changes.
• Channel navigation: List all threads in a channel or fetch unread threads across the workspace, enabling AI agents to surface relevant discussions automatically.
• Read‑state control: Mark threads as read for the current user or others, providing collaborative visibility management.

These features translate into practical use cases: an AI assistant could automatically archive outdated threads after a project ends, move urgent messages to a dedicated channel for the product team, or generate daily summaries of unread conversations. In an enterprise setting, developers can embed these tools into custom workflows that trigger on Slack commands, email notifications, or scheduled jobs—extending the reach of AI beyond a single chat interface.

Because the MCP server maps directly to Twist’s REST endpoints, developers can reason about permissions and data structures using familiar API concepts. The server’s design prioritizes clarity: each tool name follows a consistent pattern, making it easy to discover and invoke the desired operation from within an AI prompt. This predictability reduces friction when building sophisticated, stateful interactions that span multiple Twist actions.

In summary, the Twist MCP Server empowers AI assistants to act as dynamic, context‑aware collaborators within Twist workspaces. By consolidating inbox handling, thread manipulation, and read‑state management into a single MCP interface, developers can craft intelligent assistants that streamline communication workflows, reduce manual overhead, and keep teams focused on high‑value tasks.