Academy
2. Browsing the Server Catalog
Navigate the MCP server catalog to discover available services, evaluate their capabilities, and identify the right integrations for your use case.
The MCP Hub is your gateway to extending OUPI's AI with external services. MCP — the Model Context Protocol — is a standard that lets AI use third-party services as tools: reading emails, creating tasks, querying databases, and more. The Server Catalog is the dedicated section where you browse every available MCP service. Each entry in the catalog represents a server — an external service that has been packaged to speak MCP. Think of it as an app store, but instead of installing apps, you're giving your AI new capabilities. Open the catalog from the MCP Hub to start exploring what's available.
When you open a server in the catalog, you see a detail page describing what that service offers. This is where you evaluate whether a service fits your use case. Each server exposes a set of tools — discrete actions the AI can invoke. Before creating a connection, review these tools carefully: they define the exact scope of what the AI will be able to do with that service. The tool list on a connection mirrors what the AI sees at runtime, so the catalog detail page is your primary evaluation surface. No guesswork — what you see is what the AI gets.
Once you identify the right server, creating a connection is straightforward: open the service, authenticate via its authorization flow (typically OAuth), and grant the scopes it requests. That single connection then becomes available platform-wide — in ChatPro (visible as tool badges), OUPI Studio, the Assistant, workflow steps, RAG imports, and Deep Research. You don't need to reconnect per feature. This connect-once-use-everywhere model means your catalog evaluation directly determines what every AI-powered surface in OUPI can access.
When evaluating servers in the catalog, think in terms of tool granularity. A server that exposes fine-grained tools (e.g., 'read_message', 'send_message', 'list_folders') gives you more control over what the AI does than one with broad, coarse actions. Pair this with the scopes you grant during OAuth to tightly control access.
After connecting a service, immediately check the execution log. Trigger a test interaction in ChatPro and verify the tool call appears with the expected outcome. This confirms the connection is live and the tools behave as the catalog described. The log records tool name, timestamp, origin, and result.
Open the MCP Hub and navigate to the Server Catalog. Pick a service relevant to your workflow, open its detail page, and review the tools it exposes. Then create a connection, complete the OAuth flow, and verify the new tools appear in your connection's tool list. Finally, check the Execution Log after a test call.
The Server Catalog is where you discover and evaluate MCP services before connecting them. Each server lists the tools it exposes — the exact capabilities your AI gains. Creating a connection via OAuth makes those tools available across every OUPI surface: ChatPro, Studio, workflows, RAG, and Deep Research. Always inspect tools before connecting, grant only the scopes you need, and use the execution log to verify everything works. One connection, platform-wide power.