Clean raw HTML into LLM-ready text to save context and token costs.
Clean raw HTML into LAG-ready text and reduce wasted model tokens.
