local-llmdeadpixel3/claude-local-llmDelegate routine coding work to a local OpenAI-compatible model and see its live stats (time, tokens, speed, context) above the prompt