What MCP is for
MCP is a connection method for capabilities that live outside Hermes, such as repository tools, databases, or internal APIs. The official integration supports local stdio servers, remote HTTP servers, tool discovery, and per-server filtering. Start from a concrete task and expose only the relevant operations. A successful connection should be judged by an actual tool result, not by the presence of a saved server entry.
What you need
- A reachable server with documentation for its transport, authentication, runtime, and permissions.
- An appropriate Hermes profile and any local server runtime required by the chosen deployment.
Set it up
- 1
Prefer a documented or catalogued route and inspect the server's source and permission scope before connecting.
- 2
Configure the transport and credentials using the supported Hermes connection flow. Filter unnecessary tools, then begin with a read-only query against a known object.
Check it works
- Confirm authorization and tool discovery separately; an authorized account can still have unavailable tools.
- Compare the retrieved record against the original service and confirm that errors identify account, access, or connectivity gaps.
Limits and failure points
- MCP does not remove the underlying service's access restrictions, costs, or usage limits.
- A listed tool may write data. Connection approval should not be treated as blanket permission to send, delete, purchase, or publish.
Primary sources
Spotted something wrong or out of date? Tell us on X at @HermesWatcher (opens in a new tab) and we will fix it.