Provider Keys - HoneyHive AI Docs

Provider keys let HoneyHive make API calls to AI providers on your behalf. These are used for:

LLM Evaluators - Run AI-powered evaluations on your traces. See LLM Evaluators for setup.
Playground - Test prompts directly in HoneyHive

AI Provider Secrets settings page

Supported providers

Provider	Required credentials
Anthropic	API key (`api_key`)
OpenAI	API key (`api_key`)
Azure OpenAI	API key (`api_key`), endpoint (`api_base`), API version (`api_version`), deployment name (`deployment_name`)
Bedrock	Access key (`aws_access_key_id`), secret key (`aws_secret_access_key`), region (`aws_region_name`)
Gemini	API key (`api_key`)
Vertex AI	Google application credentials JSON (`google_application_credentials`), project ID (`vertexai_project`), location (`vertexai_location`)
OAuth Gateway (self-hosted only)	Gateway API base URL (`api_base`), OAuth2 token endpoint (`oauth_token_url`), client ID (`oauth_client_id`), private key PEM (`oauth_private_key_pem`), audience (`oauth_audience`). Optional: space-separated scopes (`oauth_scopes`), extra HTTP headers, one per line (`oauth_extra_headers`), JWT signing algorithm (`oauth_algorithm`: `PS256` (default), `ES256`, or `RS256`; match the algorithm your identity provider registered the client for)

Configuring provider keys

Go to Settings > Workspace > AI Providers.
Find the provider you want to use and click the pencil icon.
Enter the required credentials.
Click Save.

A green indicator and Configured status show that the required credentials have been saved for the provider.

Provider keys are configured at the workspace level. All projects within the workspace share the same provider keys.

Workspace Admins have full control over provider keys (create, read, update, delete). Workspace Members can view provider configuration status and use configured providers, but cannot view plaintext secret values or modify credentials. See Roles for details.