Description
OpenAPPA is an innovative information-flow policy engine designed specifically for LLM agents. It provides a deterministic guardrail that is completely resistant to data exfiltration caused by prompt injection or model hallucination. Unlike other solutions, OpenAPPA does not break agents and is open, vendor-agnostic, and MIT-licensed.
Read More
The engine operates by tracking data flows instead of matching patterns, ensuring that every trajectory carries a security label based on audience and trust levels. This approach allows OpenAPPA to outperform competitors on benchmarks, significantly improving task completion rates while minimizing security breaches.
OpenAPPA's configuration is declarative and data-specific, allowing it to scale to millions of agents without modification. It provides machine-readable remedy plans to help agents proceed legally when actions are blocked by policy. These plans include sanitizers to transform payloads, authorities to approve specific actions, and subagents to isolate untrusted reads.
The core concepts of OpenAPPA include security labels, tool contracts, and remedy plans. Security labels track the sensitivity and trust of data, while tool contracts define how tools interact with the agent's security state. Remedy plans offer solutions when actions do not meet tool contracts, ensuring agents remain useful under restrictions.
OpenAPPA is ideal for organizations seeking a robust and scalable solution to manage information flow and security in LLM agents. Its unique approach to deterministic security makes it a valuable tool for enhancing agent performance and safeguarding sensitive data.
- CategoryAI Agent Builders
- Pricing ModelOpen source
- DeploymentCloud
- PlatformsWeb
OpenAPPA's Core Features
Deterministic information-flow policy
Resistant to prompt injection
Vendor-agnostic and open-source
Tracks data flows with security labels
Machine-readable remedy plans
Scalable to millions of agents
Declarative configuration
Supports sanitizers and subagents
How to use OpenAPPA?
Configure: Set up security labels and tool contracts
Integrate: Plug into existing agent loops
Monitor: Track data flows and enforce policies
Optimize: Use remedy plans to maintain agent utility
OpenAPPA's Use Cases
- Data Security
- Agent Performance
- Scalable Deployments
- Policy Enforcement
- Custom Integrations
