With the fast development of distributed network technology and speeding increase of network bandwidth, it has now become one of the hottest issues as how network can be used to enrich the living quality of people. This paper will focus on the e-learning and On-line examination domain, and would like to make use diverse functions of Multi-agent as found in every Distributed Database or AP Server, and through the Data Mining technology to arrange the latest learning procedure and test recording of each learner. And after analysis, the results will be provided to the Recommending Learning AP Server as the suitable level of customization learning program suggested to learning under current level based upon system evaluation. Thus, learner can enjoy the fun of learning with sequential progress. Moreover, we have resorted to the technology of various Agents, and recorded the information of browsing and learning processes of learner in every stage, and it would then arrange and organize possible current level of learning for among learners from Database and Data Warehouse so as to provide learners to conduct judgment as they choose degree of difficulty with test-paper for online test, achieving the objective of integrate learning and test. Hence, it can efficiently improve defects of traditional e-learning and online test. At the end, we have put forth Context-aware technology to integrate intelligent online examination system so as to achieve the relevant learning content of students for extension as well as mastery of teacher for learning extent of students. Then, it can have designed the test that accommodates to student's level so that online test and learning can be complementary, helping examination to return to its original essence-appreciation of self and turning learning to be extended indefinitely.