Skip to main content

Microsoft Graph Agent MCP Server

Project description

Microsoft Agent - A2A | AG-UI | MCP

PyPI - Version MCP Server PyPI - Downloads GitHub Repo stars GitHub forks GitHub contributors PyPI - License GitHub

GitHub last commit (by committer) GitHub pull requests GitHub closed pull requests GitHub issues

GitHub top language GitHub language count GitHub repo size GitHub repo file count (file type) PyPI - Wheel PyPI - Implementation

Version: 0.9.0

Overview

Microsoft Graph MCP Server + A2A Supervisor Agent

It includes a Model Context Protocol (MCP) server that wraps the Microsoft Graph API and an out-of-the-box Agent2Agent (A2A) Supervisor Agent.

Manage your Microsoft 365 tenant (Users, Groups, Calendars, Drive, etc.) through natural language!

This repository is actively maintained - Contributions are welcome!

Capabilities:

  • Comprehensive Graph API Coverage: Access thousands of Microsoft Graph endpoints via MCP tools.
  • Supervisor-Worker Agent Architecture: A smart supervisor delegates tasks to specialized agents (e.g., Calendar Agent, User Agent).
  • Secure Authentication: Supports OAuth, OIDC, and other authentication methods.

MCP

MCP Tools

This server provides tools for a vast array of Microsoft Graph resources. Due to the large number of tools, they are organized by resource type.

Supported Resources (Partial List):

  • Users: get_user, update_user, list_user, etc.
  • Groups: get_group, post_groups_group, list_members_group, etc.
  • Calendar: get_calendar, post_events, list_calendarview, etc.
  • Drive/DriveItems: get_drive, search_driveitem, upload_driveitem, etc.
  • Mail: send_mail, list_messages, etc.
  • Directory Objects: get_directoryobject, check_member_objects, etc.
  • Planner, OneNote, Teams, and more.

All tools generally follow the naming convention: action_resource (e.g., list_user, delete_group).

Using as an MCP Server

The MCP Server can be run in two modes: stdio (for local testing) or http (for networked access).

Run in stdio mode (default):

microsoft-agent --transport "stdio"

Run in HTTP mode:

microsoft-agent --transport "http"  --host "0.0.0.0"  --port "8000"

AI Prompt:

Find who manages the 'Engineering' group and list its members.

AI Response:

I've found the 'Engineering' group.
Owners:
- Jane Doe (Head of Engineering)

Members:
- John Smith
- Alice Johnson
- Bob Williams
...

A2A Agent

This package includes a powerful A2A Supervisor Agent that orchestrates interaction with the Microsoft MCP tools.

Architecture

The system uses a Supervisor Agent that analyzes user requests and delegates them to domain-specific Child Agents.

---
config:
  layout: dagre
---
flowchart TB
 subgraph subGraph0["Agent System"]
        S["Supervisor Agent"]
        subGraph1["Specialized Agents"]
            CA["Calendar Agent"]
            GA["Group Agent"]
            UA["User Agent"]
            DA["Drive Agent"]
            OA["...Other Agents"]
        end
        B["A2A Server"]
        M["MCP Tools"]
  end
    U["User Query"] --> B
    B --> S
    S --Delegates--> CA & GA & UA & DA & OA
    CA & GA & UA & DA & OA --> M
    M --> api["Microsoft Graph API"]

     S:::agent
     B:::server
     U:::server
     CA:::worker
     GA:::worker
     UA:::worker
     DA:::worker
     OA:::worker

    classDef server fill:#f9f,stroke:#333
    classDef agent fill:#bbf,stroke:#333,stroke-width:2px
    classDef worker fill:#dcedc8,stroke:#333
    style B stroke:#000000,fill:#FFD600
    style M stroke:#000000,fill:#BBDEFB
    style U fill:#C8E6C9
    style subGraph0 fill:#FFF9C4

Component Interaction

  1. User sends a request (e.g., "Schedule a meeting with the Engineering team").
  2. Supervisor Agent identifies this as a calendar and group task.
  3. Supervisor delegates finding the group members to the Group Agent.
  4. Group Agent calls list_members_group tool and returns emails.
  5. Supervisor delegates scheduling to the Calendar Agent with the retrieved emails.
  6. Calendar Agent calls post_events tool.
  7. Supervisor confirms completion to the User.

Graph Architecture

This agent uses pydantic-graph orchestration for intelligent routing and optimal context management.

---
title: Microsoft Agent Graph Agent
---
stateDiagram-v2
  [*] --> RouterNode: User Query
  RouterNode --> DomainNode: Classified Domain
  RouterNode --> [*]: Low confidence / Error
  DomainNode --> [*]: Domain Result
  • RouterNode: A fast, lightweight LLM (e.g., nvidia/nemotron-3-super) that classifies the user's query into one of the specialized domains.
  • DomainNode: The executor node. For the selected domain, it dynamically sets environment variables to temporarily enable ONLY the tools relevant to that domain, creating a highly focused sub-agent (e.g., gpt-4o) to complete the request. This preserves LLM context and prevents tool hallucination.

Usage

MCP CLI

Short Flag Long Flag Description
-h --help Display help information
-t --transport Transport method: 'stdio', 'http', or 'sse' [legacy] (default: stdio)
-s --host Host address for HTTP transport (default: 0.0.0.0)
-p --port Port number for HTTP transport (default: 8000)
--auth-type Auth type: 'none', 'static', 'jwt', 'oauth-proxy', 'oidc-proxy' (default: none)
... (See standard FastMCP auth flags)

A2A CLI

Endpoints

  • Web UI: http://localhost:9000/ (if enabled)
  • A2A: http://localhost:9000/a2a (Discovery: /a2a/.well-known/agent.json)
  • AG-UI: http://localhost:9000/ag-ui (POST)
Argument Description Default
--host Host to bind the server to 0.0.0.0
--port Port to bind the server to 9000
--provider LLM Provider (openai, anthropic, google, huggingface) openai
--model-id LLM Model ID nvidia/nemotron-3-super
--mcp-url MCP Server URL http://microsoft-agent:8000/mcp

Examples

Run A2A Server

microsoft-agent-server --provider openai --model-id gpt-4o --api-key sk-... --mcp-url http://localhost:8000/mcp

Docker

Build

docker build -t microsoft-agent .

Run MCP Server

docker run -p 8000:8000 microsoft-agent

Run Agent Server

docker run -e CMD=agent-server -p 9000:9000 microsoft-agent

Deploy as a Service

docker pull knucklessg1/microsoft-agent:latest

docker run -d \
  --name microsoft-agent \
  -p 8000:8000 \
  -e HOST=0.0.0.0 \
  -e PORT=8000 \
  -e TRANSPORT=http \
  knucklessg1/microsoft-agent:latest

Install Python Package

python -m pip install microsoft-agent
uv pip install microsoft-agent

Repository Owners

GitHub followers GitHub User's stars

Documentation:

Microsoft API Docs Microsoft Graph SDK

MCP Configuration Examples

1. Standard IO (stdio) Deployment

{
  "mcpServers": {
    "microsoft-agent": {
      "command": "uv",
      "args": [
        "run",
        "microsoft-mcp"
      ],
      "env": {
        "ADMINTOOL": "True",
        "AGENT_DESCRIPTION": "<YOUR_AGENT_DESCRIPTION>",
        "AGENT_SYSTEM_PROMPT": "<YOUR_AGENT_SYSTEM_PROMPT>",
        "AGREEMENTSTOOL": "True",
        "APPLICATIONSTOOL": "True",
        "AUDITTOOL": "True",
        "AUTHTOOL": "True",
        "CALENDARTOOL": "True",
        "CHATTOOL": "True",
        "COMMUNICATIONSTOOL": "True",
        "CONNECTIONSTOOL": "True",
        "CONTACTSTOOL": "True",
        "DEFAULT_AGENT_NAME": "<YOUR_DEFAULT_AGENT_NAME>",
        "DEVICESTOOL": "True",
        "DIRECTORYTOOL": "True",
        "DOMAINSTOOL": "True",
        "EDUCATIONTOOL": "True",
        "EMPLOYEE_EXPERIENCETOOL": "True",
        "FILESTOOL": "True",
        "GROUPSTOOL": "True",
        "IDENTITYTOOL": "True",
        "MAILTOOL": "True",
        "METATOOL": "True",
        "MICROSOFT_ENDPOINTS_JSON": "<YOUR_MICROSOFT_ENDPOINTS_JSON>",
        "MICROSOFT_TOKEN": "<YOUR_MICROSOFT_TOKEN>",
        "MISCTOOL": "True",
        "NOTESTOOL": "True",
        "OIDC_CLIENT_ID": "<YOUR_OIDC_CLIENT_ID>",
        "ORGANIZATIONTOOL": "True",
        "PLACESTOOL": "True",
        "POLICIESTOOL": "True",
        "PRINTTOOL": "True",
        "PRIVACYTOOL": "True",
        "REPORTSTOOL": "True",
        "SEARCHTOOL": "True",
        "SECURITYTOOL": "True",
        "SITESTOOL": "True",
        "SOLUTIONSTOOL": "True",
        "STORAGETOOL": "True",
        "SUBSCRIPTIONSTOOL": "True",
        "TASKSTOOL": "True",
        "TEAMSTOOL": "True",
        "TESTING": "<YOUR_TESTING>",
        "USER": "<YOUR_USER>",
        "USERTOOL": "True"
      }
    }
  }
}

2. Streamable HTTP (SSE) Deployment

{
  "mcpServers": {
    "microsoft-agent": {
      "command": "uv",
      "args": [
        "run",
        "microsoft-mcp",
        "--transport",
        "http",
        "--host",
        "0.0.0.0",
        "--port",
        "8000"
      ],
      "env": {
        "ADMINTOOL": "True",
        "AGENT_DESCRIPTION": "<YOUR_AGENT_DESCRIPTION>",
        "AGENT_SYSTEM_PROMPT": "<YOUR_AGENT_SYSTEM_PROMPT>",
        "AGREEMENTSTOOL": "True",
        "APPLICATIONSTOOL": "True",
        "AUDITTOOL": "True",
        "AUTHTOOL": "True",
        "CALENDARTOOL": "True",
        "CHATTOOL": "True",
        "COMMUNICATIONSTOOL": "True",
        "CONNECTIONSTOOL": "True",
        "CONTACTSTOOL": "True",
        "DEFAULT_AGENT_NAME": "<YOUR_DEFAULT_AGENT_NAME>",
        "DEVICESTOOL": "True",
        "DIRECTORYTOOL": "True",
        "DOMAINSTOOL": "True",
        "EDUCATIONTOOL": "True",
        "EMPLOYEE_EXPERIENCETOOL": "True",
        "FILESTOOL": "True",
        "GROUPSTOOL": "True",
        "IDENTITYTOOL": "True",
        "MAILTOOL": "True",
        "METATOOL": "True",
        "MICROSOFT_ENDPOINTS_JSON": "<YOUR_MICROSOFT_ENDPOINTS_JSON>",
        "MICROSOFT_TOKEN": "<YOUR_MICROSOFT_TOKEN>",
        "MISCTOOL": "True",
        "NOTESTOOL": "True",
        "OIDC_CLIENT_ID": "<YOUR_OIDC_CLIENT_ID>",
        "ORGANIZATIONTOOL": "True",
        "PLACESTOOL": "True",
        "POLICIESTOOL": "True",
        "PRINTTOOL": "True",
        "PRIVACYTOOL": "True",
        "REPORTSTOOL": "True",
        "SEARCHTOOL": "True",
        "SECURITYTOOL": "True",
        "SITESTOOL": "True",
        "SOLUTIONSTOOL": "True",
        "STORAGETOOL": "True",
        "SUBSCRIPTIONSTOOL": "True",
        "TASKSTOOL": "True",
        "TEAMSTOOL": "True",
        "TESTING": "<YOUR_TESTING>",
        "USER": "<YOUR_USER>",
        "USERTOOL": "True"
      }
    }
  }
}

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

microsoft_agent-0.9.0.tar.gz (58.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

microsoft_agent-0.9.0-py3-none-any.whl (57.7 kB view details)

Uploaded Python 3

File details

Details for the file microsoft_agent-0.9.0.tar.gz.

File metadata

  • Download URL: microsoft_agent-0.9.0.tar.gz
  • Upload date:
  • Size: 58.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.3

File hashes

Hashes for microsoft_agent-0.9.0.tar.gz
Algorithm Hash digest
SHA256 7395982b02a30598f2c344f8a48ae9268e7e401a2471b7b864be11c90bfce251
MD5 00c9e410903b032791a4369e40112ed8
BLAKE2b-256 6ec5c3d677208632e9ba7716cfe17365dae9a65973c7e9d7d5b786c310eab92d

See more details on using hashes here.

File details

Details for the file microsoft_agent-0.9.0-py3-none-any.whl.

File metadata

File hashes

Hashes for microsoft_agent-0.9.0-py3-none-any.whl
Algorithm Hash digest
SHA256 06db35d8199755cb983202bfab3110c8c830f96cbadf18cb60a05a398751d0d7
MD5 3c11795d8fa2ac3c5c5cd16515c76aca
BLAKE2b-256 9b819ae4d7e10b0c2d05d43e9b272a103f2542b46bef439e2e58192b4d3ec8e9

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page