Webcmd (by AgentR) is open-source Released on October 8/9, 2026 (markets.businessinsider.com)

AgentR has launched Webcmd, a free open-source browser infrastructure that enables AI agents to learn and remember websites for more efficient task execution. Released on October 8/9, 2026, Webcmd allows agents to build a locally stored sitemap that records their interactions with a site, including pages, workflows, and APIs. This innovation significantly reduces the need for agents to repeat tasks from scratch, cutting token costs by up to 90%. In a benchmark test, Webcmd completed 67 tasks with an average cost of $0.255 per task, outperforming competing tools and showcasing its potential for cost-effective browser automation. This release is significant for the AI/ML community as it represents a shift towards self-learning systems that adapt based on real-world usage rather than static training data. With Webcmd’s functionality, developers can issue combined JavaScript commands, trimming unnecessary data and ensuring tasks can continue even with memory failures. The project has quickly gained traction, becoming the top trending JavaScript repository shortly after its launch, reflecting strong interest in creating more efficient AI tools. By making Webcmd open-source, AgentR enables developers to inspect and collaborate on this new foundation for browser automation, fostering further innovation in the field.