Connects a self-hosted OmniRoute gateway so agents can see its model catalog, routing decisions, per-provider health, cost and usage. Changing routing or flushing caches needs your approval. Needs a running OmniRoute instance (eve add omniroute) and an inference key from its dashboard.
What your agent can do with it
- omniroute_mcpGoverned gateway to the OmniRoute MCP server. The pod routes an agent's call to a non-inline tool (set_routing_strategy, cache_flush, switch_combo) through this row, where the cap…
- omniroute-gateway-advisorWhen and how to use the OmniRoute gateway's MCP tools. Load when a model/provider choice depends on what the gateway can serve, a route behaves unexpectedly, or the user asks abou…
What you will need
Asked for at install. Stored in your pod, never on our servers.
- Gateway URLBase URL of the OmniRoute gateway WITHOUT a trailing path, reachable from the intelligence service — e.g. https://omniroute.example.com. The MCP endpoint is <b…
- Inference API keyAn inference key (sk-…) from OmniRoute's dashboard → API Keys. Stored server-encrypted as a pod-wide shared key; used only as the MCP transport's Authorization…
- OmniRoute inference keySecretService credential for the OmniRoute gateway's /api/mcp endpoint.