POST /mltk/vector_stores/test

Description. Probe an AWS Knowledge Base. Three input modes:

• Saved — {"name": "kb-production"}.
• Inline — full type + details.
• Hybrid — name plus a partial details where some fields are ***MASKED***. Masked values are pulled from saved credentials.

Behavior. Uses boto3.client('bedrock-agent-runtime').retrieve(...) with test query {"text": "connectivity test"} and numberOfResults: 2.

Response (200).

{
  "connected": true,
  "status":    "success",
  "message":   "Successfully connected to Knowledge Base EJGJR4SKJL in us-west-2",
  "details": {
    "kb_id":        "EJGJR4SKJL",
    "region":       "us-west-2",
    "role_arn":     "arn:aws:iam::123456789012:role/SplunkMCPRole",
    "test_results": 2,
    "validation":   "Full KB retrieve test passed"
  }
}

Common errors. ResourceNotFoundException (wrong kb_id/region), AccessDeniedException (IAM), or "boto3 version doesn't support Bedrock services" (the bundled boto3 in older Splunk needs an upgrade).

Example.

curl -sk -u "$SPLUNK_USER:$SPLUNK_PASSWORD" -X POST 
  -H "Content-Type: application/json" 
  --data '{"name":"kb-production"}' 
  "$SPLUNK_HOST/servicesNS/nobody/Splunk_ML_Toolkit/mltk/vector_stores/test?output_mode=json"

Source. bin/rest_handlers/vector_stores.py, bin/connection_config_manager/vector_db/config_manager.py.