Show HN: Agentshield Zero-leak kernel sandbox and secret defense for AI coding

Category: security

Tags: security, ai-agents, sandbox, secrets-management, devsecops

Score: 6.5/10 (Innovation: 6, Technical: 6, Documentation: 8, Utility: 6)

AgentShield is a TypeScript-based security layer that sandboxes AI coding agents and redacts secrets to prevent credential exfiltration into LLM contexts or Git history. It combines OS-level sandboxing (Seatbelt, bubblewrap, AppContainer), AST-based command interception, and streaming token redaction, with integrations for pre-commit hooks and GitHub Actions.

Target audience: backend devs, devops, security engineers, AI tooling developers

Repository: https://ritualdev.in/agentshield/ · TypeScript · Apache-2.0 · 3 stars

View on Hacker News