MakerChecker: Testing AI agents for risky behavior before deployment

MakerChecker is an open-source tool that audits AI agents for dangerous capabilities—like unauthorized API access or data exfiltration—before they go live. For indie makers building with LLMs, this addresses a real gap: most agent frameworks lack built-in safety checks, and shipping blind is a liability.