Registering and monitoring MCP servers
Registering an MCP server puts it under G8KEPR's watch: every tool it exposes is inventoried, pinned, and checked on each call, and the server's health is monitored automatically. This guide shows how to register a server and keep an eye on its tools and status from the dashboard.
Before you begin
- ▸An account with access to the MCP Security area of the dashboard.
- ▸The connection details for your MCP server (its URL or command, and any auth it requires).
- ▸If you route traffic through G8KEPR, make sure your client points at G8KEPR rather than the server directly. See Connect your first API key.
Register a server
- 1.Go to MCP Security → Servers in the dashboard.
- 2.Click Register server and give it a recognizable name.
- 3.Enter the server's endpoint (URL for a remote server, or the launch command for a local one) and any required credentials.
- 4.Save. G8KEPR connects, enumerates the server's tools, and records a canonical fingerprint of each one.
You can also register programmatically:
{
"name": "billing-mcp",
"endpoint": "https://mcp.internal.example.com",
"auth": { "type": "bearer", "token": "..." }
}Register in a staging environment first so you can review the tool inventory before any live traffic flows through the server.
Monitor tools
Open a server from the registry to see its Tools tab, which lists every tool the server advertises along with its current fingerprint. When you approve a tool, G8KEPR pins that fingerprint and re-checks it on every call. If a tool's definition later changes, the call is flagged and can be blocked automatically.
A silently changed tool definition is a classic rug-pull. To learn how these checks work, read How rug-pull and tool-poisoning detection works.
Monitor health and activity
- ▸Servers shows each server's connection status and last-seen time, so you can spot one that has gone offline or unreachable.
- ▸Sessions tracks active connections and the tool calls flowing through them.
- ▸Threats surfaces anything blocked or flagged, such as a fingerprint mismatch or a suspicious tool call.
Health checks run automatically once a server is registered. Set up notifications so your team is alerted when a server goes down or a tool changes unexpectedly.