A fixed-sample observatory of twelve public transgender-history source groups in mainland China, comparing government HTML, institutional landing pages, PDFs, community archives, and legal mirrors, and proposing a machine-readable evidence ladder and AI discovery debt model.