Academic libraries need to connect knowledge resources and support services with teaching agents,but existing studies provide limited guidance on how these contributions can be organized across teaching workflows.This study aims to clarify the functional layers and implementation pathways through which academic libraries can participate in agent-supported teaching.It reviewed research on teaching agents,library teaching support,and intelligent library services,and identified five practical foundations:resource advantages,platform infrastructure,application contexts,collaboration mechanisms,and role redefinition.A framework comprising resources,technology,services,collaboration,and evaluation was then developed and applied to Northwestern Polytechnical University Library.The case analysis aligned each layer with library practices and examined questionnaire responses alongside service data.The questionnaire used stratified multistage sampling and obtained 780 valid responses from 1,000 distributed questionnaires,yielding a valid response rate of 78%.Descriptive statistics,group comparisons,and regression analysis were used to examine service use,perceived benefits,and associated factors.The framework distinguished the provision of teaching tools and subject resources from application deployment,platform interfaces,and data analysis.It connected these foundations to teaching design,course support,tool training,learning spaces,and AI information literacy,while assigning coordination and feedback responsibilities to the collaboration and evaluation layers.The survey showed that 85%of respondents used intelligent service platforms frequently or occasionally,80%perceived improved learning and research efficiency,and 65%had participated in AI information literacy activities.Overall satisfaction with intelligent service platforms reached 93%.Use frequency and training participation were positively associated with perceived service effectiveness.These findings suggest that embedding teaching agents requires coordinated strategic planning,resource and technology integration,service development,and improvement of librarians'AI capabilities.Libraries should define responsibilities jointly with teaching and technical departments and use evaluation results to refine services.The study offers a framework linking library responsibilities to concrete teaching tasks and provides a reference for context-sensitive implementation.However,the single-institution design,nonrandomized departmental comparisons,and concurrent educational reforms limit causal attribution and generalization.Further longitudinal and cross-institutional research is needed to determine the independent contribution of teaching agents and the transferability of the proposed pathways.