Skip to content

Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.

fix: show configured VLM in status without usage (#3148)

huangruiteng378173705b0e
diff --git a/openviking/storage/observers/models_observer.py b/openviking/storage/observers/models_observer.py
index 3c5736d60..4786073f6 100644
--- a/openviking/storage/observers/models_observer.py
+++ b/openviking/storage/observers/models_observer.py
@@ -61,12 +61,15 @@ class ModelsObserver(BaseObserver):
# VLM section
if self._vlm_instance:
+ vlm_data = None
try:
vlm_data = self._get_vlm_usage()
- if vlm_data:
- sections.append(("VLM", vlm_data))
except Exception as e:
logger.warning(f"Error getting VLM usage: {e}")
+ if not vlm_data:
+ vlm_data = self._get_configured_vlm()
+ if vlm_data:
+ sections.append(("VLM", vlm_data))
# Embedding section
if self._embedding_instance:
@@ -125,6 +128,20 @@ class ModelsObserver(BaseObserver):
return data
+ def _get_configured_vlm(self) -> Optional[list]:
+ """Return configured VLM identity when usage data is unavailable."""
+ model = getattr(self._vlm_instance, "model", None)
+ if not model:
+ return None
+
+ return [
+ {
+ "Model": model,
+ "Provider": getattr(self._vlm_instance, "provider", None) or "unknown",
+ "Status": "configured",
+ }
+ ]
+
def _get_embedding_usage(self) -> Optional[list]:
"""Get Embedding token usage data."""
if not self._embedding_instance:
diff --git a/tests/misc/test_models_observer.py b/tests/misc/test_models_observer.py
new file mode 100644
index 000000000..d1df8e348
--- /dev/null
+++ b/tests/misc/test_models_observer.py
@@ -0,0 +1,34 @@
+# Copyright (c) 2026 Beijing Volcano Engine Technology Co., Ltd.
+# SPDX-License-Identifier: AGPL-3.0
+"""Focused tests for model status observability."""
+
+from openviking.storage.observers.models_observer import ModelsObserver
+
+
+class _ConfiguredVLM:
+ model = "astron-code-latest"
+ provider = "litellm"
+
+ def get_token_usage(self):
+ return {"usage_by_model": {}}
+
+
+class _ConfiguredVLMWithUnavailableUsage(_ConfiguredVLM):
+ def get_token_usage(self):
+ raise RuntimeError("usage backend unavailable")
+
+
+def test_configured_vlm_is_visible_before_usage_is_recorded():
+ status = ModelsObserver(vlm_instance=_ConfiguredVLM()).get_status_table()
+
+ assert "VLM Models:" in status
+ assert "astron-code-latest" in status
+ assert "litellm" in status
+ assert "configured" in status
+
+
+def test_configured_vlm_is_visible_when_usage_lookup_fails():
+ status = ModelsObserver(vlm_instance=_ConfiguredVLMWithUnavailableUsage()).get_status_table()
+
+ assert "VLM Models:" in status
+ assert "astron-code-latest" in status