JBAssist Microsoft Graph MCP Server

JBAssist – Microsoft Graph MCP Server

JBAssist bridges Claude Desktop with the full breadth of Microsoft 365 through the Model Context Protocol. By exposing Graph API endpoints as MCP tools, it lets AI assistants perform real‑world actions—such as reading emails, pulling calendar events, or querying user directories—without the developer having to write custom connectors. This eliminates a common pain point for teams looking to embed AI into their existing Office 365 workflows: the need to manage OAuth, permissions, and API calls manually.

The server bundles both standard and beta Graph capabilities. Standard tools cover the essentials: , , , and . These enable Claude to fetch user details, inbox snapshots, upcoming meetings, or search the organization directory. The beta suite extends this with richer data—, , , and —and advanced filtering via . This depth is especially valuable for HR, support desks, or collaboration bots that need to surface nuanced employee information.

Key features include:

• Native MCP integration – Claude can invoke any tool directly from a conversation, passing natural language prompts that the server translates into Graph queries.
• Fine‑grained permission handling – The setup process guides developers through Azure AD app registration and API consent, ensuring the server has only the scopes it needs.
• Cross‑platform Windows support – A lightweight Node.js server runs on Windows 10/11, making it easy to deploy in corporate environments that already run Microsoft 365.
• Beta API access – By exposing beta endpoints, JBAssist gives developers early access to new Graph features without waiting for official releases.

Typical use cases span from personal productivity (auto‑reading unread emails or scheduling meetings) to enterprise automation (retrieving a manager’s direct reports for onboarding workflows). In a customer support scenario, an AI agent could pull the caller’s calendar and presence status to suggest optimal meeting times. HR bots can use to find candidates with specific skills across the organization.

Integrating JBAssist into an AI workflow is straightforward: once the MCP server is running, Claude Desktop’s puzzle‑piece selector adds “msgraph” as a tool source. From there, developers can craft prompts that trigger Graph calls—e.g., “Show me my next three meetings” or “Find all users in the Sales department with Python skills”—and the AI responds with live data. This tight coupling between natural language and enterprise APIs dramatically reduces friction for developers, allowing them to focus on business logic rather than authentication plumbing.