DocSpry
DocSpry reads whole books, reports, theses and journal articles the way a production editor does, and gives Claude the result as data it can quote: chapters with their headings, paragraphs in reading order, lists, block quotes, tables (as Markdown or HTML), figures with their captions and images, display equations with LaTeX where it can be recognised, footnotes linked to their references, and the table of contents. Every block carries the page it came from and its position on that page, so Claude can cite a passage by page number and a reviewer can find it in the original. Typical use: upload a 400-page book once, then ask Claude to summarise chapter 7 with page citations, pull every table from the methods section, extract the bibliography as structured records, search for where a term is defined, or compare two chapters. Reading a converted document is free and unlimited; the document stays in the user's DocSpry account and can be opened in the DocSpry editor for a human review pass. DocSpry also publishes: one call renders the same document to reflowable EPUB, fixed-layout EPUB, accessible PDF (tagged, PDF/UA), JATS, BITS, DocBook or Elsevier XML, HTML, chapter JSON, or a Markdown bundle, with the validator's report alongside the download link. Converting a document is metered per page on the user's DocSpry balance (the connector shows the cost before starting, and converting the same document again is free). DocSpry converts born-digital PDFs and Word files; scanned PDFs are declined and never charged. A quality report accompanies every conversion so Claude can say how far to trust a page and where a person should look.
Details
https://api.docspry.com/mcp · streamable-http
