Connect via MCP
How to connect an assistant such as Claude to your workspace, what it can do once connected, and which AI actually builds the course.
Create a connection in your workspace settings, add the connection URL to your assistant, and sign in once to authorise it. From then on the assistant can create and manage courses in that workspace, using over 50 tools. Course generation itself runs on Coassemble's own AI, so a course built through your assistant behaves the same as one you build in the app.
Connect an assistant
- Open Settings, then MCP under the Developer section.
- Create a connection. Coassemble gives you a connection URL.
- Add that URL to your assistant as a remote MCP server.
- Sign in once when the assistant prompts you, to authorise it against your workspace.
The connection is scoped to the workspace you authorised it from. You can connect more than one assistant, and remove any of them without affecting the others.
What your assistant does, and what Coassemble does
Your assistant drives Coassemble. It decides what to ask for and calls the right tools, using over 50 tools covering course creation, editing, publishing and analytics. When it asks for a course to be generated, that generation runs on Coassemble's AI.
The practical effect is that you get to work in the assistant you prefer, while the quality and behaviour of a generated course stays consistent whichever way you asked for it.
Which assistants can I connect?
Any assistant that supports connecting to a remote MCP server and signing in. Because MCP is an open standard rather than a private integration, this is not a fixed list, and support depends on the assistant rather than on Coassemble.
Driving the agent from your own product
If you deliver Coassemble inside your own application, there's a second path. Your backend mints an agent token from your workspace API key and your application drives the Coassemble agent over MCP or REST, so your users get the agent without leaving your product. You can pin a token to a single course, allowlist which tools it may use, or make it read-only.
That differs from the connection above, which is one person authorising their own assistant against their own workspace.
What is MCP?
MCP, the Model Context Protocol, is an open standard for connecting AI assistants to external tools and data. It means an assistant can work with Coassemble through a documented, permissioned connection rather than a one-off custom integration built for a single product.
Frequently asked questions
Can I choose which AI model generates my course content?
No. Course generation runs on Coassemble's AI whichever assistant you connect. What you choose is the assistant you work in and the model it runs on.
Does connecting an assistant replace Coassemble's built-in AI?
No. Coassemble's built-in AI features work exactly as they do now. A connected assistant is an additional way to reach them.
Can I connect more than one assistant?
Yes. Each one appears separately in your list of authorised connections, and you can remove any of them without affecting the others.
Can an assistant read my results as well as build?
Yes. The tool set covers analytics, so a connected assistant can read course and screen level data and answer questions about it.