Skip to content

Installation

Install Sondera Harness with your preferred package manager. Choose the extra that matches your agent framework.

Requirements

  • Python 3.12 or higher
  • uv (recommended) or pip

Don't have uv? Install it: 

curl -LsSf https://astral.sh/uv/install.sh | sh

Install

Pick your framework, then run one of the commands:

uv (recommended): 

uv add "sondera-harness[langgraph]"
pip: 
pip install "sondera-harness[langgraph]"

uv (recommended): 

uv add "sondera-harness[adk]"
pip: 
pip install "sondera-harness[adk]"

uv (recommended): 

uv add "sondera-harness[strands]"
pip: 
pip install "sondera-harness[strands]"

Adds AI-powered queries to the TUI dashboard. Supports any LLM provider (Gemini, OpenAI, Anthropic, Ollama, and more).

uv (recommended): 

uv add "sondera-harness[ai]"
pip: 
pip install "sondera-harness[ai]"

uv (recommended): 

uv add sondera-harness
pip: 
pip install sondera-harness

Verify Installation

Confirm the package installed correctly:

uv run python -c "from sondera import CedarPolicyHarness; print('Sondera Harness installed successfully!')"

python -c "from sondera import CedarPolicyHarness; print('Sondera Harness installed successfully!')"

Configuration (Platform Only)

If you're using Sondera Platform (SonderaRemoteHarness), set your API credentials:

export SONDERA_HARNESS_ENDPOINT="your-harness.sondera.ai:443"
export SONDERA_API_TOKEN="<YOUR_API_TOKEN>"

Or create a .env file or ~/.sondera/env:

SONDERA_HARNESS_ENDPOINT=your-harness.sondera.ai:443
SONDERA_API_TOKEN=<YOUR_API_TOKEN>

Local evaluation doesn't need this

The Quickstart uses CedarPolicyHarness for local policy evaluation, which doesn't require API credentials. See Deployment for details on choosing local vs platform.

Next Steps