Local-first AI security scanner detects risks in agent workflows
AI agents with tool access pose serious security risks. A new open-source tool scans workflows for prompt injection, data exposure, and autonomy flaws—all locally, without cloud dependencies.