Gemini

Configure HolmesGPT to use Google's Gemini models via Google AI Studio.

Setup

Get your API key from Google AI Studio.

Configuration

export GEMINI_API_KEY="your-gemini-api-key"
holmes ask "what pods are failing?" --model="gemini/<your-gemini-model>"

Using CLI Parameters

You can also pass the API key directly as a command-line parameter:

holmes ask "what pods are failing?" --model="gemini/<your-gemini-model>" --api-key="your-api-key"

Gemini-Specific Configuration

Gemini models require special handling for tools without parameters. Set the following environment variable when using Gemini:

export TOOL_SCHEMA_NO_PARAM_OBJECT_IF_NO_PARAMS=true

This ensures that tool schemas are formatted correctly for Gemini's requirements. See the Environment Variables Reference for more details.

Additional Resources

HolmesGPT uses the LiteLLM API to support Gemini provider. Refer to LiteLLM Gemini docs for more details.