Self-Coding Repository – Idea Update

The author is developing a self-coding project using Large Language Models (LLMs), such as GPT3.5/4-turbo and GPT4, to generate code, tests, and documentation. Existing project files are used as LLM prompt context. Evaluation tools from the Python ecosystem, GitHub API and function calling options are leveraged to streamline the process. The system has been designed to cater for various programming tasks. However, tweaking different agent "personalities" and effectively managing complex project directories remain challenges.