An MCP (Model Context Protocol) server gives the AI access to external tools — a standardized way for an AI model to call out to a tool or data source beyond its own training, rather than every integration needing custom, one-off wiring.
Three transport types
- stdio — the server runs as a local process; the AI talks to it over standard input/output.
- SSE (Server-Sent Events) — a remote server the AI connects to over a persistent event stream.
- Streamable HTTP — a remote server reached over regular HTTP requests.
Health and tool count
Each connected server reports whether it's currently healthy and how many tools it exposes — a quick way to confirm a server is actually reachable and doing something, not just configured.
Related articles
Other guides that answer questions close to this one.
How to Add an MCP Server
Connect an MCP server so the AI can use its tools. See MCP Servers Explained first. Before you start The Manage AI Settings right. Steps 1. Go to Settings → Artificial Intelligence → MCP Servers. 2. Click Add Server. 3.…
How to Create an AI Access Policy
Define usage limits and provider/model access for a group of roles. Before you start The Manage AI Settings right. Steps 1. Go to Settings → Artificial Intelligence → Security Models. 2. Click Add Policy. 3. Fill in: -…
How to Add an AI Provider
Connect an AI provider so the organization's AI features have a model to call. Before you start The Manage AI Settings right, and your organization must have AI Support enabled. Steps 1. Go to Settings → Artificial…