Semantic Markdown Splitter

This tool splits markdown text into semantic chunks grouped by markdown headers (H1-H6) to optimize retrieval use cases. It accepts markdown_text (required string containing the markdown content to split), max_chunk_size (optional integer specifying a soft limit on chunk character count, defaulting to 1000 when omitted), and include_headers (optional boolean that controls whether each chunk includes the originating header text, defaulting to true when omitted). The worker must transform these inputs into JSON output that represents the resulting semantic chunks and strictly validates against the system-provided structured output schema (auto-produced, not supplied here), using correct types and field shapes and without introducing extra fields or unsupported values.