No vague success-rate, blanket privacy or safe-batch slogans. Tests are pinned to production versions, while processing-boundary and batch-safety evidence state what the current implementation actually does and what has not been verified.
Publishes 10 source-backed rename and image-batch safeguards: old/new-name preview, blocking collisions and unsafe names, original-file preservation, source-image thumbnails and deterministic output de-duplication.
Publishes local-vs-cloud boundaries for renaming, image, PDF, CSV/JSON, duplicate detection, text-to-document and OCR workflows, with High Accuracy OCR explicitly identified as the current production cloud exception.
All 25 compatibility assertions passed against the production parser module, with speed, browser-matrix and maximum-file testing explicitly excluded.
All 49 assertions passed against the production document-export module, covering normalization, safe filenames, background rules, and TXT/Markdown/HTML output; DOCX packaging and PDF rendering are explicitly out of scope.