Report

Getting one

Open Report in the sidebar, under Forecast. The first time, press Write the report. After that the Report is written again every time a Quick check runs, and Regenerate writes it from the latest runs whenever you like. If runs have happened since it was written, the Report says so at the top.

What is in it

At the top, four answers at a glance: What changed since earlier runs, What is open and how much of it is ship-blocking, what is Likely fixed, and the Next moves. Then the document itself:

• Summary. The verdict (Ready to ship, Almost there, Not ready or Not measured yet), the launch score when one has been measured, and the headline figures. On a paid plan it also gets a short written review; without one, it shows the figures only.
• How much of the product this covers. A crawl of your site from its front door, against the pages the runs opened and the forms no goal uses. Every other figure in the Report is measured against this.
• What to do next, ranked by how strongly the evidence supports each finding and how much it costs to leave it.
• State of play. Every check that has run, its latest score, and how that moved since its previous one.
• Likely fixed: verify and close. Findings that stopped appearing when later runs opened their page again. This suggests a fix but does not prove one, so each one waits to be verified.
• Open items, most severe first, and anything open more than 14 days.
• Scope and method. How the document was made, and where it draws the line between what was measured and what was judged.

Every finding in it is marked measured (a browser or a detector established it), judged (a read on the product that no browser could settle) or unchecked (a factual claim nothing has checked yet).

Sending it

• Print, or choose Save as PDF in the print dialog.
• Download text for a doc, a wiki or an email.
• Full report downloads the whole document as a PDF to keep, file or forward. It lists every finding in a register under an ID that stays the same in later reports, shows where the findings are page by page, cites the standard each one fails where one applies (WCAG and EN 301 549 for accessibility, OWASP ASVS for security, never as a claim that the product conforms), puts the picture that proves a finding beside it, and says what is fixed and what is still open. Every page after the cover names the product and the reference and says which page of how many it is, and the PDF can be read by a screen reader. If a PDF cannot be made at that moment, the same report arrives as a web page you can open and print.
• Spreadsheet downloads every finding in the register as a CSV, for Excel, Google Sheets or a tracker such as Jira.
• Share makes a read-only link, with the PDF and the Markdown beside it. A link lasts up to 1 day on Free, 7 days on Pro, and 30 days on Team and Agency. On Agency, shared reports can carry your agency’s own name.

Every Report carries the date it was issued and a reference built from the run that produced it, so two people holding the same document can tell they are reading the same one. The Readiness page has a report of its own, with its four answers and a verdict, and its own Share this report link.

How gaps are shown

A number nobody measured is printed as a dash, never as zero. If the crawl of your site fails, no coverage figure is published at all. A page no run has opened is listed as never opened, not counted as clean. If your site cannot be reached when the Report is written, the Report is built from the runs already on file and says so at the top.

A clean Report over three pages of a thirty-page product describes only those three pages. The coverage section shows how much of the product was checked.

Free on every plan

The Report needs no AI to be written, so it is free and unlimited on every plan, and it runs as part of the free Quick check.

When a run has AI, the Report also opens with a short written review of the week, drawn from your finding titles and each check’s latest score. Every number on it is still counted without AI, and the free Quick check never adds the review.

From the command line

If you run nth Labs yourself from the repository:

node cli.cjs brief       # write the Report
node cli.cjs report      # the full report as one HTML file, and its register as a CSV
node cli.cjs report --pdf   # the same, plus the PDF
node cli.cjs coverage    # the coverage section on its own

report writes nothing when there is nothing measured to put in it, instead of producing an empty document. With an AI key set on your machine, brief adds a short written review on top of the figures.