Short-term automation solutions for industrial companies have potential to compensate for volatile demand, employee turnover or shortages. Current solutions are either not available at short notice or economically unviable, as integration in existing control systems and machine tools would be necessary. The paper proposes a cognitive robotic system that circumvents this problem by emulating the behaviour of operators through automated, cognitive skills. The approach uses a perception system, handling and skill modules and a skill-based control to automate in an industrial brownfield use case without additional interfaces to the existing control systems to reduce integration effort to a minimum.