safe-agents

Training (hopefully) safe agents in gridworlds

Safe production-grade engineering skills for AI coding agents.

This is an example implementation of a Spring Boot AI Chat Agent that uses the open-policy-ai-advisor-spring-boot-starter to enforce governance and policy-driven guardrails. By integrating Open Policy Agent (OPA), this agent can evaluate prompts and tool calls against custom Rego policies before they ever reach the LLM.

Experimental safety envelope for agent-initiated mutations to existing commercial state.

Open Policy AI Advisor is a Spring Boot starter that integrates Open Policy Agent (OPA) for policy evaluation and enforcement in your applications. This starter simplifies the process of adding open policy-based advisors capabilities to your AI chat client.