This article describes Model Context Protocol (MCP) in order to help you understand:
- what MCP is;
- how it's used by Higher Logic Vanilla; and
- how you can use it in your Vanilla community.
MCP in brief
MCP stands for Model Context Protocol. It is a standard way for an AI assistant to connect to another service's tools. Vanilla MCP is the bridge between your chosen AI assistant and your Vanilla community: the assistant requests an approved tool, Vanilla returns information or carries out an action the signed-in user is allowed to perform, and the assistant uses that result to answer you. MCP is a connector, not an AI model or a separate copy of your community database.
What is MCP?
The Model Context Protocol (MCP) is an open-standard "universal connector" through which AI applications can securely connect to external tools, data sources, and services through a common interface in order to access information and perform actions without the need for a custom integration for every system (e.g., ChatGPT, Claude, Cursor, and compatible custom agents).
Vanilla MCP
Vanilla MCP is Higher Logic Vanilla's support for the Model Context Protocol to support your ability to directly and securely connect to your Vanilla community. With this secure connection, you can:
- retrieve content,
- answer questions, and
- take approved actions on live community data.
Is MCP AI?
No.
- MCP is not AI and
- Vanilla MCP does not add AI features on its own.
MCP is a connection layer that lets you securely connect your own AI tool or large language model (LLM) to Vanilla via the existing Vanilla API.
In other words, Vanilla MCP helps your chosen AI tool retrieve content from Vanilla.
Vanilla MCP is not:
- an AI model,
- a chatbot, or
- a database.
Does Vanilla MCP expose new data or permissions?
No. As a connector — not an AI model or a database — Vanilla MCP is a pass-through to your existing Vanilla API, with no new data exposed beyond what you already access in Vanilla today.
NOTE: What users can see and do remains governed by their existing Vanilla roles, permissions, and access controls.
How do I sign in?
Each person must complete the Vanilla MCP authentication page when connecting their AI tool. In the current beta, this uses a separate Vanilla email/password sign-in rather than your community's SAML SSO page. A first-time user may need to create that sign-in and verify an emailed code. SSO support is planned for general availability; it is not part of the current beta. Each person should connect with their own account so their existing Vanilla roles and permissions apply. Do not use a shared account.
How is the connection protected?
Vanilla MCP uses OAuth 2.0 for individual authentication and TLS to protect data in transit. It calls Vanilla's existing API on behalf of the signed-in user, and Vanilla applies that person's roles and permissions. The MCP server retrieves community content as needed and does not retain that content after responding to a request. Actions are logged for audit. Your chosen AI provider has its own data-handling settings and policies. For more about authentication, access controls, and data handling, see Vanilla MCP security and data handling (Beta).
What problem does it solve, and what can our team do?
Vanilla MCP replaces the manual efforts of copying threads, exporting reports, and rebuilding context. Teams use plain-language prompts to ask about, for example:
- community activity,
- content performance,
- product feedback,
- moderation queues,
- unresolved questions, and
- member behavior.
Teams can then act on what they've learned, such as follow up with members, organize content, send reminders, and automate recurring workflows when the AI tool supports those actions.
How can I learn more about Vanilla MCP?
Have a question about the connector, supported AI clients, permissions, or beta sign-in? Read Vanilla MCP: customer FAQ (Beta).
To connect ChatGPT, follow Connect Vanilla MCP to ChatGPT (Beta).
To connect Cursor, follow Connect Vanilla MCP to Cursor (Beta).
For authentication, access controls, and data handling, read Vanilla MCP security and data handling (Beta).
Join the conversation!
The Vanilla MCP beta program is underway and the community is getting engaged. Interested in joining us?
Check out some of these discussions…
Starter Use Cases
This discussion has some prompts that are meant to help you test the kinds of workflows MCP is designed to support:
Introduction to Higher Logic Vanilla MCP beta!
This discussion introduces Vanilla MCP and touches on the concerns of security and governance. Other community users are asking questions and expressing their interest to participate:
Configure MCP with Claude discussion
This discussion describes how Vanilla MCP beta users can configure MCP in their Claude environment: