Skip to content

Agent Mesh

docs: only MCP and A2A backends are probed before they are advertised

Tomas Tormoc42019884a5d
diff --git a/site/content/docs/concepts/networking.md b/site/content/docs/concepts/networking.md
index 54d575ce..ab280dfb 100644
--- a/site/content/docs/concepts/networking.md
+++ b/site/content/docs/concepts/networking.md
@@ -65,10 +65,11 @@ reprovide interval of the mesh (`/info`), withdraws a record when it
stops the service or leaves, and a record a provider stopped announcing
without withdrawing it leaves the directory after three intervals.
-Before a node advertises a backend, it probes the backend
+Before a node advertises an MCP or A2A service, it probes the backend
(`--backend-probe-timeout`, 2 seconds by default). A backend that does not
-answer is not advertised, so `discover_remote_services` does not list
-services that would fail on the first call.
+answer is not advertised, so `discover_remote_services` does not list an
+MCP or A2A service that would fail. The node advertises inference and
+egress services without a probe.
Discovery results are hints. They tell the caller who claims to provide a
service and, if the provider declared labels, what those labels are. Nothing
diff --git a/site/content/docs/reference/agentmesh-node.md b/site/content/docs/reference/agentmesh-node.md
index c2849832..80a112ae 100644
--- a/site/content/docs/reference/agentmesh-node.md
+++ b/site/content/docs/reference/agentmesh-node.md
@@ -118,7 +118,7 @@ explains how to enroll.
| `--discovery-interval` | `30s` | Delay before the first announcement of this node's services to the directory, capped at 5 s; services are re-announced every reprovide interval of the mesh after that (`agentmesh-control-plane --dht-reprovide-interval`, 5 minutes by default). |
| `--dht-lookup-limit` | `20` | Providers a directory lookup asks for. |
| `--discovery-concurrency` | `10` | Concurrent catalog fetches during discovery. |
-| `--backend-probe-timeout` | `2s` | How long a service backend may take to answer before the node declines to advertise it. Raise it for subprocesses that start slowly. |
+| `--backend-probe-timeout` | `2s` | How long an MCP or A2A backend may take to answer before the node declines to advertise it. Raise it for subprocesses that start slowly. |
| `--secrets-dir` | `/etc/agentmesh/secrets` | Directory holding the credentials that the control plane's [egress destinations](../policy/#egress-destinations) name, one file per credential name. The platform puts the files there; the node reads a file on every request to the destination. |
| `--control-plane-sync-interval` | `15m` | How often signing keys, the ban set, router addresses and the mesh policy are pulled from the control plane. Keep it well below the control plane's `--key-grace-period`; raise it on large meshes. |
| `--key-grace-period` | `24h` | How long a rotated-out control plane key is still accepted for verifying peers. |
diff --git a/site/content/docs/reference/node-config.md b/site/content/docs/reference/node-config.md
index f22fb329..2e6e4025 100644
--- a/site/content/docs/reference/node-config.md
+++ b/site/content/docs/reference/node-config.md
@@ -95,9 +95,10 @@ Rules by type:
`/.well-known/agent-card.json` on the backend and served again with mesh
URLs. The card must use the A2A 1.0 format.
-Before a service is advertised, the node probes the backend
+Before an MCP or A2A service is advertised, the node probes the backend
(`--backend-probe-timeout`, 2 seconds). A backend that does not answer is
-not advertised, and the log records this.
+not advertised, and the log records this. An inference service is
+advertised without a probe.
Services exist only through this file and through the control plane's
egress assignments. There is no runtime API that adds a service. An agent