Not your LLM, not your AI agent I guess!

Llama works locally with a VS Code plugin. Slow unless you have some decent compute though.