Connect Mindsmith to an AI tool with MCP
Mindsmith provides a remote Model Context Protocol (MCP) server. A compatible AI tool can use it to find Mindsmith content and inspect results. With your permission, it can also create or change lessons.
Server address
Use the address for the region where you sign in:
Region | MCP server |
US |
|
EU |
|
In your AI tool, add a remote MCP server and paste the appropriate address. The tool opens Mindsmith in your browser to complete authorization.
Useful setup destinations include Claude connector settings, Microsoft Copilot, and Copilot Studio.
Authorize a workspace
On the Mindsmith authorization screen:
Sign in to Mindsmith.
Select the workspace the assistant may access.
Leave Write access off for a read-only connection, or turn it on if the assistant should create and change content.
Select Authorize Access.
Read access is always granted. The connection is limited to the selected workspace, and every action remains subject to your Mindsmith role and access to the affected content.
Use an organization API key
An organization Admin can also connect a client by sending a Mindsmith organization API key as a bearer token. Use the same regional MCP server address shown above.
API-key MCP connections are read-only. They are not tied to a real Mindsmith user, so MCP does not allow them to create or change content. Use the browser-based OAuth flow when a connected tool needs write access.
The full organization key is shown only when it is created. Store it securely, and revoke it under Settings and Members → API keys when the client should no longer have access.
What a connected assistant can do
With read access, a connected tool can:
List projects, courses, and lessons
Search lesson content
Inspect a lesson’s structure and sharing status
Inspect a course’s ordered lesson outline
Read top-level lesson results and tile-level engagement
Drill into one learner’s engagement when a learner identifier is provided
Authors, Admins, and organization API keys can also inspect images in the organization Media Library.
With write access, a connected tool can:
Create lessons from instructions
Edit existing lessons
Publish or unpublish lessons
Generate individually tracked learner links
Create standalone AI conversation lessons when that feature is available
Lesson creation and editing run asynchronously. The assistant may need to check the lesson again after generation finishes.
What data passes through the connection
Mindsmith doesn’t connect to your files, drives, or other systems. Your AI tool decides what to send to Mindsmith, and the tool’s own access controls apply before anything is sent.
Your AI tool can send Mindsmith:
Instructions and lesson content as text
Learner names and email addresses, when it creates tracked links
The connection doesn’t accept file uploads. To build from a document or video, attach it to the lesson or course in Mindsmith first. An assistant with write access can then work from the attached file.
Mindsmith can return to your AI tool:
Lesson and course content and structure
Results and engagement, including each learner’s name, email address, answers, scores, and time
Images from the organization Media Library, for Authors, Admins, and organization API keys
Every role in the selected workspace, including Viewer, can read results and engagement through a connection. An organization API key can read them across the organization. Once Mindsmith returns data to your AI tool, that tool’s own terms and data handling apply to it.
Creating or editing a lesson from instructions uses Mindsmith’s AI, under the same commitments as the Mindsmith Agent: your content isn’t used to train AI models, and AI providers don’t retain it. See Security & Data Privacy for details.
Each server address connects to one region, so a connection to the EU server reads and writes content and learner data held in the EU.
Example requests
"List the courses in the Onboarding project."
"Find the lesson about incident reporting and summarize its structure."
"Show the completion rate and the tiles where learners disengage most."
"Create a lesson called Quarterly Security Refresher in the Compliance project."
"Publish this lesson and create tracked links for these learners."
Be explicit about the workspace, project, lesson, audience, and desired outcome when names could be ambiguous.
Sharing and tracked links
A published lesson’s public learner URL supports broad distribution. Activity is recorded, and learners can identify themselves when the lesson is configured to collect names or email addresses. To associate a unique link with a specific learner in advance, ask the assistant for tracked links instead. Tracked links work for published or unpublished lessons and keep each learner’s activity separate in analytics.
Change or revoke access
If a client loses authorization, remove or disconnect the Mindsmith server in that client and connect it again. During reauthorization you can choose a different workspace or change whether write access is granted.
Disconnect the integration from the AI tool when it should no longer access Mindsmith. Organization offboarding also revokes user-bound MCP access for that organization.