CAPSEM is a comprehensive framework designed to enhance the privacy and security of AI agents by allowing to define, manage, and enforce contextual privacy and security policies in a framework-agnostic manner.
Important Links: Docs,
-
[Oct 25] PII Security Policy: New policy that can block, confirm, or log based on detected PII types in model and tools responses.
-
[Oct 25 ] Transparent Proxying: Seamlessly proxy requests through CAPSEM proxy to transparently enforce policies on industry leading API based models including OpenAI and Google Gemini.
-
[Sept 25] ADK Support: Initial release with ADK support.
-
Framework Agnostic: CAPSEM is designed to be framework-agnostic, allowing easy integration with various AI agent frameworks.
-
Policy Management: Define and manage complex privacy and security policies that can be tailored to specific use cases.
-
Contextual Policies: Define and enforce privacy and security policies based on the context of the agent's operations.
-
Transparent Proxying: Proxy requests through CAPSEM to enforce policies on external models.
-
Extensible Architecture: Easily extend CAPSEM with custom policies and integrations.
This project is licensed under the Apache 2.0 License - see the LICENSE file for details.
We welcome contributions from the community! Whether it's bug reports, feature requests, documentation improvements, or code contributions, please see Code Contributing Guidelines to get started.
This is not an officially supported Google product. This project is not eligible for the Google Open Source Software Vulnerability Rewards Program.
