Infisical released Agent Vault, an open-source credential proxy that prevents AI agents from directly accessing secrets. Rather than returning credentials, it intercepts HTTP requests and injects the required credentials at the network layer, eliminating exfiltration risks from prompt injection. The tool works with any HTTP-capable agent, including Claude Code and Cursor, with AES-256-GCM encryption.
Safety
Show HN: Agent Vault – Open-source credential proxy and vault for agents
Agent Vault intercepts HTTP requests to inject credentials at the network layer, preventing AI agents from handling secrets directly and blocking prompt-injection exfiltration attacks.
Friday, April 24, 2026 12:00 PM UTC2 MIN READSOURCE: Hacker NewsBY sys://pipeline
Tags
safety