AgentLeash

Hey Product Hunt 👋 I'm Sky, solo founder of VERIDIAN. I use AI coding agents daily—Claude Code, Cursor, Windsurf. They're transformative. But one thing kept bothering me: I had no idea what they were actually accessing. These agents read your entire codebase to build context. They can touch config files, environment references, lockfiles. And you just... trust that it's fine. So I built AgentLeash. It's file access monitoring designed specifically for AI agents: $ leash init $ leash watch Now I see every file operation in real-time. I define boundaries in a simple YAML config—block .env files, restrict writes to certain directories, set different rules for different agents. When an agent tries to access something restricted: [X] READ .env.local [BLOCKED] I get notified. I see the attempt logged. I stay in control. The free tier gives you 1 scope and 100 logs/day—enough to try it on a real project. I'd love to hear: - What files do you wish you could protect from agent access? - Any specific agent integrations you'd want? Happy to go deep on implementation or roadmap. — Sky