Description
When configuring OpenKB to use Google Cloud Vertex AI (vertex_ai/gemini-3.8-flash) using standard GCP Application Default Credentials (ADC) rather than a static API key, I get the below error:
1. PageIndexError: API key not configured for provider 'vertex_ai'
- What happens: Running
openkb add or initializing PageIndexClient triggers _validate_llm_provider in pageindex, which enforces a static VERTEX_AI_API_KEY env var.
- Expected behavior: Cloud providers that authenticate via ADC / IAM (
vertex_ai) should be recognized as credentialed without demanding a static API key.
- Additionally, OpenKB's CLI prints a false
Warning: No LLM API key found for these providers.
We have resolved this locally by:
Hooking PageIndexClient._validate_llm_provider to allow ADC/IAM cloud providers.
Updating _KEYLESS_OR_CLOUD_PROVIDERS in cli.py to prevent false missing-key warnings.
Description
When configuring OpenKB to use Google Cloud Vertex AI (
vertex_ai/gemini-3.8-flash) using standard GCP Application Default Credentials (ADC) rather than a static API key, I get the below error:1.
PageIndexError: API key not configured for provider 'vertex_ai'openkb addor initializingPageIndexClienttriggers_validate_llm_providerinpageindex, which enforces a staticVERTEX_AI_API_KEYenv var.vertex_ai) should be recognized as credentialed without demanding a static API key.Warning: No LLM API key foundfor these providers.We have resolved this locally by:
Hooking PageIndexClient._validate_llm_provider to allow ADC/IAM cloud providers.
Updating _KEYLESS_OR_CLOUD_PROVIDERS in cli.py to prevent false missing-key warnings.