Skip to main content
Glama
Sign In
msteams-mcp

msteams-mcp

by anielyavuz
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Local

msteams-mcp

A small, self-hosted MCP server for Microsoft Teams that lets an AI assistant (VS Code Copilot Chat, Claude, …) read your own Teams chats — with least privilege, on your machine, under your own Entra app.

  • Delegated only — acts as the signed-in user; sees only what that user sees (/me/...).

  • Read-only — v0.1 has 4 read tools; the Graph client is GET-only.

  • Your own app registration — single tenant, no secret, no application permissions, assigned users only.

  • Local — HTTP on 127.0.0.1 with a bearer token (or stdio). No telemetry.

  • Auditable — one log line per request (account, client, tool, result) without message content.

  • Extensible — new Teams capabilities are added as separate tools with declared scopes (roadmap).

⚠️ Data leaves your machine through your AI client. Tool results — including Teams message text — are sent to the AI model your MCP client uses (usually a cloud service). Review docs/DATA-FLOW.md before using it with organizational data.

Tools (v0.1)

Tool

What it returns

Scope

auth_status

Which account the server acts as, granted scopes, token expiry

–

get_current_user

Your profile

User.Read

list_chats

Your chats (1:1, group, meeting), members, last message preview; filter by type / text

Chat.Read

get_chat_messages

Messages of one of your chats as plain text; limit, time window, sender filter, order

Chat.Read

Related MCP server: Teams MCP

Quick start

Requirements: Node.js 20+, an Entra app registration (docs/ENTRA-SETUP.md).

git clone https://github.com/anielyavuz/msteamsMCP.git && cd msteamsMCP
npm ci --ignore-scripts
cp .env.example .env && chmod 600 .env      # set TENANT_ID, CLIENT_ID, MCP_ACCESS_TOKEN
npm run login                               # device code sign-in
npm start                                   # http://127.0.0.1:3978/mcp
npm run selftest -- --live                  # second terminal: security + live checks

Then add the server to your MCP client (docs/OPERATIONS.md).

How it works

AI client ──HTTP 127.0.0.1 + Bearer──▶ msteams-mcp ──HTTPS, delegated token──▶ Microsoft Graph (/me/chats…)
    │                                       │
    └──▶ AI model provider (cloud)          └── token cache 0600, audit log (no content)

Details: docs/ARCHITECTURE.md (modules, functions, flows, adding a tool).

Documentation

Document

For

docs/DATA-FLOW.md

Security/privacy review: what data goes where, incl. the AI provider

SECURITY.md

Controls, limitations, reporting vulnerabilities

docs/ENTRA-SETUP.md

App registration, permissions, restricting users

docs/OPERATIONS.md

Commands, configuration, clients, logs, troubleshooting

docs/ARCHITECTURE.md

Components, functions, request and sign-in flows

docs/ROADMAP.md

Planned tools and the review rule for each

docs/KURULUM.tr.md

Adım adım kurulum (Türkçe)

License

MIT © 2026 Anıl Yavuz. Dependencies: @modelcontextprotocol/sdk, @azure/msal-node, zod (all MIT).

Not affiliated with or endorsed by Microsoft. Microsoft, Microsoft Teams, Microsoft Entra and Microsoft Graph are trademarks of the Microsoft group of companies.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    This server cannot be deployed

    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    Resources

    Related MCP Connectors

    Related MCP Servers