The rapidly growing collection of clinical free text data about cancer provides an unprecedented opportunity to harness rich, real-world insights for advancing cancer research and treatment. At the same time, there is a formidable challenge in efficiently extracting, standardizing, and structuring this vast and unstructured information to derive meaningful conclusions. In this paper, we address the problem of information extraction to standardize and structure such free text using Large Language Models (LLMs) following the minimal Common Oncology Data Elements (mCODE™) structure. We propose mCodeGPT, a state-of-the-art data standardization tool Cancer related clinical notes. mCodeGPT ingests Cancer ontology knowledge base, automatically generates prompts in a hierarchical manner to extract information within the ontology from the clinical notes, and outputs results in a structured, tabular format. The pipeline is completely zero-shot, meaning it is annotation-free and does not require model training. Our new tool simplifies the conventional labor-intensive process of data collection, annotating, model training, and fine-tuning. To overcome challenges inherent in common LLMs, such as token size limitations, hallucination, and suboptimal accuracy in information extraction, we have developed three distinct methodologies for hierarchical prompt engineering. We conducted rigorous evaluations of mCodeGPT's efficacy in the domain of Cancer clinical note generation and information extraction. Employing the GPT-3.5-turbo model, we generated a diverse set of 400 simulated clinical notes spanning various Cancer types. The effectiveness of our three proposed prompt generation methods was further assessed using both OpenAI models and a selection of open-source LLM models. The outcomes of our research underscore the proficiency of our pipeline, showcasing an impressive 96% accuracy in information extraction. We published mCodeGPT on GitHub for other researchers to use.