Install & Configure

The Giskard scan needs two things before it can run: the Python package, and an LLM provider with an API key. The scan cannot work without a model, because a model is what writes the attacks and grades the replies.

Install with a coding agent

The fastest way to set up the scan. Paste a single URL into your coding agent and it handles the install and the provider configuration for you.

Install the Python package

The scan requires Python 3.12 or higher:

pip install --pre "giskard[scan]"

uv pip install --prerelease=allow "giskard[scan]"

Configure a model

The scan calls an LLM twice: once to invent attack scenarios from your description, and once to judge the answers your agent gives back, meaning to read each conversation and decide pass or fail. Both use the default generator, so nothing runs until you set one.

Those calls send your agent’s description and its replies to whichever provider you configure. If your agent can return customer data or internal documents, that data reaches the provider too.

Install the provider extra alongside the scan:

pip install --pre "giskard[scan,openai]"

Then register the model as the default:

from giskard.agents.generators import GiskardLLMGenerator
from giskard.checks import set_default_generator

set_default_generator(GiskardLLMGenerator(model="openai/gpt-4o"))

openai, anthropic, and google are the first-party extras. For anything else, install the litellm extra and pass any LiteLLM-supported ↗ model string, such as "mistral/mistral-large-latest", "azure/gpt-4o", or "ollama/llama3".

Each provider reads its own API key from the environment (OPENAI_API_KEY, ANTHROPIC_API_KEY, GEMINI_API_KEY). Keep them in a .env file and load it before you build the generator:

pip install python-dotenv

from dotenv import load_dotenv

load_dotenv()

Pick a capable model here. The generator writes the attacks and the judge decides whether your agent fell for them. A weak model produces bland scenarios and unreliable verdicts, which is the most common cause of a scan that “finds nothing”.

Optional: third-party scanners

third_party_scan runs external red-teaming tools in-process through lazily imported adapters. They are not installed by default:

pip install --pre "giskard[garak]"
pip install --pre "giskard[deepteam]"

Install only the ones you plan to run; both pull in large dependency trees. See third_party_scan for the arguments each tool accepts. A probe there is one canned attack that tool knows how to run, the equivalent of a Giskard generator.

Next Steps

Start with Your First Scan for a guided run against a toy agent, or go straight to Scan Vulnerabilities to point the scan at your own.