Oghmere docsOpen the app
Help & docs

Integrity Report

An exportable, shareable record of how a document was written — evidence of process, not a claim of innocence.

The Integrity Report is a per-document record of how your work was written: that it grew over time under your hand, that the AI in the loop asked questions and gave feedback but never wrote prose, and that the sources you cited exist. It is built for the moment a student needs to answer the modern question — not “will I be caught”, but “how do I prove I wrote this?” — and it is the one answer Oghmere can give credibly, because Oghmere never generated any of the text in the first place.

What it is — and is not

It evidences process: a record of how the document was written in Oghmere. It is not an AI-detector verdict and not a certificate of innocence — see “What it proves” below. It reads telemetry you already have; it runs no AI and spends no credits.

What the report shows

  • A headerthe document title, author, project, date range, and total elapsed writing time.
  • The statementa plain statement that Oghmere never generates prose and that every word was typed by its author, linked to the public charter that explains how that is enforced.
  • A draft timelinea chart of word count over calendar time with version markers — organic growth over days or weeks looks nothing like a single paste-in, and reads at a glance.
  • A revision summaryhow many versions, over what span, and the largest single addition between saves. That large-addition signal is shown plainly, never hidden — the report is trustworthy precisely because it does not smooth over anything.
  • An AI interaction logeach coaching run: which coach, when, what class of output (a question, a measurement, or feedback), and whether you acted on it — never the content of the coaching, which stays your private work.
  • A source ledgerevery cited source with its DOI or ISBN where known, so a reader can check that the references exist.

How to generate one

Reports are off until you ask for one. Generate a report from the document you want it for — nothing is recorded or shared until you do. Generating a report does not publish it: it is yours to review, export, and share only if you choose to.

  • Export itdownload the report as a PDF to attach to a submission.
  • Share itcreate a private, unguessable link that renders the same report from a frozen snapshot — so the figures a reader sees do not change after you shared them. The link shows only what the report shows; it never exposes your document’s text.
  • Revoke itturn off a shared link at any time, and it stops working immediately. Deleting the document removes its reports and invalidates any links you shared.
  • An independent timestampwhere one is available, a shared report is timestamped by an independent timestamp authority — not by Oghmere. A reader can check that timestamp without us, and it shows the figures have not changed since the report was issued. It says nothing about who wrote the document, or how; that is not something any timestamp can show.

What it proves — and what it does not

The report evidences process, not innocence. It shows how the document was written in Oghmere, and an instructor draws their own conclusion. It cannot prove that text was not drafted somewhere else and pasted in — which is exactly why a large single addition is shown rather than hidden. It is deliberately not a cryptographic certificate: a signature would prove that Oghmere issued the report, not that you wrote the words. This restraint is the point. Claiming more certainty than the record supports would put the report in the same position as the AI detectors — confidently wrong — and that is the one thing it must never do.

The charter behind the statement

The report’s statement links to Oghmere’s no-generation charter — a public page at /charter that explains why Oghmere never generates prose and how the coaching constraint tests enforce it on every push. That charter is what the whole report rests on.

What it costs

Nothing. The report reads telemetry already captured as you wrote — draft versions, coaching runs, cited sources — and adds no AI of its own, so it spends no credits.