The Robust Intelligence AI Firewall is a model-agnostic, external guardrail that secures LLM-powered applications from prompt injection, PII extraction, and other harmful actions and content. Informed by our proprietary threat intelligence, it validates model inputs and outputs in real-time and can be configured to block or modify undesired responses.
This notebook is a tutorial for using the AI Firewall Python SDK. For complete setup and documentation of the Firewall, please see the user guide.
The legacy Firewall SDK is included in the Robust Intelligence RIME SDK. You can install it via pip as shown here:
pip install rime-sdk==2.10.16
Note: If you run into issues, check to see if there is a newer 2.10 patch version to install here: https://pypi.org/project/rime-sdk/
See Firewall SDK and REST API, above, for more information about the SDK.