Show HN: A Sidecar-Based Policy Enforcement Engine for AI Agents

1 point

a year ago

I’ve prototyped sidecar-based policy enforcement engine as a proof of concept to help AI agents (like CrewAI or LangGraph) comply with organizational rules. It intercepts requests/responses, applies YAML-defined policies (e.g., disallowed phrases/tools), and can use an LLM for semantic checks. I’m looking for feedback, suggestions, or collaborators—please check it out - explains the core idea in README and let me know what you think!