Developing domain-specific frameworks for large language model (LLM) applications has been widely studied to enhance research efficiency across various fields. As policy planning requires detailed analysis of reliable sources, LLM applications have significant potential to enhance analytical efficiency. Therefore, we propose a novel LLM application framework designed for policy planning that reduces the time and resources required for analyzing global policy trends and their comprehensive content, including goals, implementation processes, outcomes, and causal relationships. First, a systematic collection and evaluation of datasets to identify reliable sources of global policy information that could serve as valuable references for policymakers was conducted. This process involved indicator development for data quality assessment in terms of reliability, relevance, completeness, and timeliness. We then categorized types of policy planning based on an in-depth analysis of previous policy documents, establishing key content requirements for each category. Subsequently, the fine-tuning and retrieval-augmented generation (RAG) approaches for the datasets were suggested and validated to efficiently extract required information from the collected datasets. We expect the framework to help policymakers effectively analyze global policy landscapes, identify policy implementation strategies along with their results, and derive valuable insights for policy planning across diverse contexts and challenges.