Show HN: Stop Destructive Agent Commands Before They Happen

Category: security

Tags: ai-safety, policy-enforcement, cli-tool

Score: 7.0/10 (Innovation: 7, Technical: 7, Documentation: 6, Utility: 8)

Orca is a policy enforcement layer for autonomous AI agents, intercepting risky actions like destructive commands, file access, and network calls before execution. It fills a clear gap in AI safety by providing reusable, cross-agent guardrails without requiring Docker or VMs.

Target audience: backend devs, devops, security engineers

Repository: https://github.com/christopherkarani/Orca · Rust · Apache-2.0 · 27 stars

View on Hacker News