An MCP (Model Context Protocol) server implementation for Microsoft Teams integration that can read/create/reply-to messages and mention members in Teams channels.
https://github.com/InditexTech/mcp-teams-serverStop alt-tabbing between Claude and Teams. This MCP server connects your AI tools directly to Microsoft Teams, letting you read messages, create threads, reply to conversations, and mention team members without leaving your chat interface.
You're building something in Claude, need to check what the team discussed, so you switch to Teams. Then you want to share an update, so you copy-paste between apps. Need to mention someone? Another context switch. It's death by a thousand tab switches.
The MCP Teams Server fixes this by making Teams channels and conversations available directly in your AI tools. Read team context, post updates, and manage conversations from wherever you're already working.
Read Team Context: Pull recent messages from any channel to give your AI full context about ongoing discussions, decisions, and blockers.
Post Without Switching: Create new discussion threads or reply to existing ones directly from Claude, ChatGPT, or any MCP-compatible tool.
Smart Mentions: Tag specific team members in your AI-generated messages - perfect for routing questions to the right people or escalating issues.
Thread Management: Start focused discussions with titles and content, then follow up with threaded replies to keep conversations organized.
Daily Standups: Ask your AI to summarize yesterday's team discussions, then post today's priorities directly to the standup channel with relevant team mentions.
Code Review Coordination: Pull context from your dev channel about a PR discussion, generate a summary of the technical decisions, and post it back with mentions for stakeholders who need to sign off.
Customer Issue Escalation: Read the support channel for context about a customer issue, draft a detailed technical response, and post it with mentions for the customer success team.
Sprint Planning: Analyze team discussions about upcoming features, generate user stories or technical specs, and share them directly in your planning channel.
Built on Microsoft Graph APIs with proper enterprise authentication through Azure AD. Supports both single-tenant and multi-tenant configurations, so it works with your organization's existing Teams setup.
The server runs locally or in Docker, connecting to your Teams instance through standard Microsoft Bot Framework protocols. Set your app credentials, point it at your team and channel, and you're reading and writing Teams messages through MCP.
# Quick setup
git clone https://github.com/InditexTech/mcp-teams-server.git
cd mcp-teams-server
uv sync --frozen --all-extras
# Configure your Teams app (see their setup guide)
export TEAMS_APP_ID=your-app-id
export TEAMS_APP_PASSWORD=your-client-secret
export TEAM_ID=your-team-id
export TEAMS_CHANNEL_ID=your-channel-id
# Start the server
uv run mcp-teams-server
The hardest part is the initial Azure AD app setup (they provide a detailed guide), but once configured, the server just works. No complex webhooks or callback URLs - straightforward environment variables and you're connected.
InditexTech (the folks behind Zara) built this for their own enterprise workflows, so it handles the authentication complexity and error cases you'd expect in a corporate environment. Docker support, comprehensive error handling, and integration tests included.
At 259 stars and growing, it's already being used by teams who needed this exact workflow improvement. The codebase is clean, well-documented, and actively maintained.
Your AI tools should work with your existing team communication patterns, not against them. This MCP server makes that happen.