I got tired of AI that thinks for me – so I created AkitaLLM (github.com)

AkitaLLM has been launched as a command-line utility that aims to redefine how developers interact with AI by prioritizing structured engineering processes over generative capabilities. Unlike traditional AI tools that function in a "black-box" mode, AkitaLLM enforces a disciplined workflow of Analyze, Plan, Execute, and Validate, allowing developers to maintain control over their projects. This approach is particularly significant as it addresses the shortcomings of existing AI applications that often fail to understand project-level context or to ensure code integrity before making changes. The tool operates as a local-first application, maintaining strong security standards by keeping user code on their machines while orchestrating models and APIs through encrypted channels. It utilizes recursive file scanning to create a comprehensive map of the project context, which informs its suggestions. With features like automated local test execution, detailed structural code reviews, and the ability to generate actionable diffs for manual review, AkitaLLM positions itself as an essential component in the modern software engineering process, promoting accountability and rigorous oversight in AI-assisted coding. Available on PyPI, it invites developers to contribute to its evolution, emphasizing the importance of robust abstractions and clean code in AI-driven development.