France-based startup Edamame has launched a runtime security platform designed to detect when AI coding agents drift from their intended purpose and potentially cause security breaches. The system monitors agents like Cursor and Claude Desktop through six integrated layers that track divergence from developer intent and detect attack patterns such as credential harvesting and token exfiltration, while also catching supply-chain attacks that reach developer workstations through compromised packages. The platform aims to shift the security question from trusting developers to verifying that AI agents stay within their operator's original instructions during execution.