LitLLM is a toolkit designed to assist in the literature review process for scientific papers. It uses Retrieval Augmented Generation (RAG) principles, specialized prompting, and LLMs to generate accurate and relevant literature reviews. The system first summarizes user-provided abstracts into keywords for web search, then retrieves and re-ranks relevant papers based on the abstract. The related work section is generated from the re-ranked results, significantly reducing the time and effort required for traditional literature reviews. The toolkit is open-source and available on GitHub and Hugging Face. The system allows users to provide an abstract or research idea, which is then used to search for relevant papers. Users can enhance the search by adding relevant keywords or papers. The retrieved papers are re-ranked based on the user's abstract, and the related work section is generated using this information. The system also supports sentence-based planning to ensure controllable and tailored generation of the literature review. LitLLM addresses the limitations of existing LLM-based literature review tools, such as hallucination and outdated knowledge. By incorporating RAG techniques, the system ensures factual accuracy and avoids generating non-factual information. The modular pipeline allows for easy adaptation to new LLMs and other domains, such as news. The toolkit has been tested with five researchers, who found the zero-shot generation informative and the plan-based generation more accessible for their research. The system's ability to generate concise and usable literature reviews makes it a valuable tool for researchers. Future work includes exploring academic search through multiple APIs and leveraging longer context LLMs to ingest full papers for more relevant background information.LitLLM is a toolkit designed to assist in the literature review process for scientific papers. It uses Retrieval Augmented Generation (RAG) principles, specialized prompting, and LLMs to generate accurate and relevant literature reviews. The system first summarizes user-provided abstracts into keywords for web search, then retrieves and re-ranks relevant papers based on the abstract. The related work section is generated from the re-ranked results, significantly reducing the time and effort required for traditional literature reviews. The toolkit is open-source and available on GitHub and Hugging Face. The system allows users to provide an abstract or research idea, which is then used to search for relevant papers. Users can enhance the search by adding relevant keywords or papers. The retrieved papers are re-ranked based on the user's abstract, and the related work section is generated using this information. The system also supports sentence-based planning to ensure controllable and tailored generation of the literature review. LitLLM addresses the limitations of existing LLM-based literature review tools, such as hallucination and outdated knowledge. By incorporating RAG techniques, the system ensures factual accuracy and avoids generating non-factual information. The modular pipeline allows for easy adaptation to new LLMs and other domains, such as news. The toolkit has been tested with five researchers, who found the zero-shot generation informative and the plan-based generation more accessible for their research. The system's ability to generate concise and usable literature reviews makes it a valuable tool for researchers. Future work includes exploring academic search through multiple APIs and leveraging longer context LLMs to ingest full papers for more relevant background information.