I built VibeLint because AI coding tools made me faster… but also more nervous.

An agent can write and edit files for you, one bad suggestion can ship a hardcoded secret, missing auth check, unsafe query, or exposed env variable before you even notice.

VibeLint is an open-source security layer for AI coding workflows. It connects through MCP and scans AI-generated code before it gets written, flagging risky patterns in real time.

The goal is simple: keep the speed of vibe coding, but add a safety net before mistakes reach production.

Would love feedback from other founders and devs building with AI agents.