Minimalist framework for language-guided robot navigation using frozen vision-language embeddings. Achieves 74% success rate without fine-tuning. RSS 2025 Workshop paper. - View it on GitHub