At present, the operating procedures of nuclear power plants mainly rely on word software for writing and circulation. There are differences in the compatibility of software versions, and the industry historical data cannot be effectively used for intelligent prompts. To solve these problems, this paper focuses on the design and application of general implementation software for the digitization of nuclear power regulations. With the help of QT framework, a visual interface is built, and an adaptive parsing strategy is used to balance the parsing efficiency. Semantic understanding is realized through word embedding and context coding, location aware coding and dynamic memory network, and intelligent prompt and automatic completion algorithms are designed, including prompt generation based on reinforcement learning, sequence to sequence automatic completion, data retrieval and result mapping, to provide efficient intelligent support for nuclear power regulation editing.