When you expand a server, MCPFlo triggers capability discovery — a round-trip to the server asking it to list its tools, resources, and prompts. Most servers respond quickly, but a slow, unresponsive, or misbehaving server can leave that request hanging.

Cancelling a slow fetch

If a server’s discovery is taking too long:

1. Locate the server row (or its loading indicator) in the sidebar.
2. Use the cancel action to abort the in-flight discovery request.
3. The server row returns to a non-connected state, ready to retry.

This means one slow or stuck server doesn’t block you from browsing or working with your other connected servers — the discovery request is scoped per-server and cancelling it doesn’t affect anything else.

When you’d use this

• A server hangs during development — common when you’re actively iterating on a server and it’s temporarily in a broken state; cancel and retry once you’ve fixed it, without restarting MCPFlo.
• A remote Streamable HTTP server is unreachable — rather than waiting out a long timeout, cancel immediately and check connectivity.
• You expanded the wrong server — cancel before it finishes if you didn’t mean to trigger discovery.

After cancelling

• Re-expand the server row to retry discovery from scratch.
• If it continues to hang, see Troubleshooting Connections for connection-specific diagnostics (stdio process issues, HTTP timeouts, etc.).
• You can also raise the connection timeout in the server’s Advanced settings if the server is simply slow rather than stuck — see Protocol Version Overrides for where the Advanced section lives.