Automated test tools are an essential resource for practitioners responsible for evaluating the accessibility of Web sites. However, both systematic analysis of tool capabilities and practitioner feedback have identified a range of practical issues that mar the effectiveness of existing tools. in practice, although automated test tools need to be used in combination to give good coverage, their lack of consistent user experience and their diverse reporting formats discourage such combined usage. Furthermore, test tools are expensive to develop; in addition to core analytical capability, authors must individually construct the user interface, I/O routines, Web crawlers, and report writers. In this paper, an architecture is proposed to address these concerns. in this architecture, tools are developed as plug-ins to an infrastructure that provides a common user interface, crawling and parsing services, and practitioner-oriented tools for analysis and reporting. The architecture supports an efficient, systematic evaluation process and benefits accessibility practice in two distinct ways: first, it simplifies the task of the evaluator by providing a consistent, integrated, and efficient user experience for executing, reporting, and communicating a study; second, it supports an economic model in which tools can release development resources from mundane software engineering activities in order to invest in the intelligent-agent development necessary to address the deeper challenges of automated testing.