diff --git a/README.md b/README.md
index 5c4b5cadf..d28c4a870 100644
--- a/README.md
+++ b/README.md
@@ -220,7 +220,7 @@ During onboarding, NemoClaw validates the selected provider and model before it
 
 Credentials stay on the host in `~/.nemoclaw/credentials.json`. The sandbox only sees the routed `inference.local` endpoint, not your raw provider key.
 
-Local Ollama is supported in the standard onboarding flow. Local vLLM remains experimental, and local host-routed inference on macOS still depends on OpenShell host-routing support in addition to the local service itself being reachable on the host.
+Local Ollama is supported in the standard onboarding flow. On Linux Docker hosts, Ollama must listen on `0.0.0.0:11434` so sandboxes can reach `http://host.openshell.internal:11434`; a loopback-only `127.0.0.1:11434` bind will fail validation. Local vLLM remains experimental, and local host-routed inference on macOS still depends on OpenShell host-routing support in addition to the local service itself being reachable on the host.
 
 ---
 
diff --git a/docs/reference/inference-profiles.md b/docs/reference/inference-profiles.md
index f1c1a4f49..1b6b32d9d 100644
--- a/docs/reference/inference-profiles.md
+++ b/docs/reference/inference-profiles.md
@@ -76,6 +76,16 @@ Ollama gets additional onboarding help:
 - it warms the model
 - it validates the model before continuing
 
+On Linux hosts that run NemoClaw with Docker, the sandbox reaches Ollama through `http://host.openshell.internal:11434`, not the host shell's `localhost` socket.
+If Ollama is already running, make sure it listens on `0.0.0.0:11434` instead of `127.0.0.1:11434`.
+For example:
+
+```console
+$ OLLAMA_HOST=0.0.0.0:11434 ollama serve
+```
+
+If Ollama only binds loopback, NemoClaw can detect it on the host but the sandbox-side validation step fails because containers cannot reach it.
+
 ## Experimental Local Providers
 
 The following local providers require `NEMOCLAW_EXPERIMENTAL=1`: