fix(cli): silence LiteLLM 'could not pre-load' warnings at import time

LiteLLM unconditionally tries to pre-load AWS Bedrock / SageMaker
response stream shapes during 'import litellm'. When 'botocore'
isn't installed it logs two WARNING lines per invocation — but
botocore is optional and OpenAI / Anthropic / Gemini users have no
reason to install it.

Result was that every 'openkb' call printed two unhelpful warnings
above the actual output, including for terminal/agent consumers
parsing the first line of 'openkb status' to get the KB path.

Attach a 'logging.Filter' to the 'LiteLLM' logger BEFORE litellm
imports, dropping any record whose message contains 'could not
pre-load'. Real LiteLLM warnings still come through.

Author: KylinMountain

openkb/cli.py

@@ -23,6 +23,18 @@
 os.environ.setdefault("LITELLM_LOCAL_MODEL_COST_MAP", "True")
 
 import click
+
+# Silence LiteLLM's "could not pre-load <aws-service> response stream
+# shape" warnings — they fire at import time when ``botocore`` isn't
+# installed, but botocore is only needed for AWS Bedrock / SageMaker
+# users. Filter must be attached before ``import litellm`` runs.
+class _SuppressLiteLLMPreloadWarnings(logging.Filter):
+    def filter(self, record: logging.LogRecord) -> bool:
+        return "could not pre-load" not in record.getMessage()
+
+
+logging.getLogger("LiteLLM").addFilter(_SuppressLiteLLMPreloadWarnings())
+
 import litellm
 litellm.suppress_debug_info = True
 from dotenv import load_dotenv