Nexus StudioHow does Nexus Studio check the numbers on this site?
Every figure on this site comes from a command we run or a test in our repository, or it is marked as an estimate. This page lists the main ones, where each comes from, the check, and what the check printed on the date shown. It also says what the checks do not prove.
By the Nexus Studio team, Verdict System LLC. October 27, 2026.
Which claims have a check, and what did it print?
| What we say | Where it comes from | The check we run | Result on October 27, 2026 |
|---|---|---|---|
| There are 14 verdicts, and 7 of them exit 0 | classification-module.js, the list of passing verdicts and exitCodeOf | node --test tests/unit/classification-parity-engines.test.js | 42 tests, 42 passed |
| A failure is tested against the environment first, in a fixed order, before it can be a bug | The same file, the test "detection order: the first matching family wins, on both engines" | The same command | Part of the 42 |
| The Studio and the engine group the verdicts the same way | verdict-groups.js and the Studio's status.ts | node --test tests/unit/verdict-groups.test.js | 6 tests, 6 passed |
| A changed API response is reported as API_DRIFTED, with the difference | api-schema.js, and the example on the schema drift page, run by hand | node --test tests/unit/api-schema.test.js | 10 tests, 10 passed |
| No command reaches a shell, and no command is built by string interpolation | no-shell.test.js, which reads the source | node --test tests/unit/no-shell.test.js | 3 tests, 3 passed |
| The release gate has 6,057 tests: 2,997 engine, 3,060 Studio | npm run gate: the engine tests, the Studio type check, the Studio tests and lint with zero warnings | npm run gate | 0 failed and 9 skipped (7 in the engine, 2 in the Studio), lint with 0 warnings |
| What the site says about the product, such as the AI tools it drives, its timeouts and the Playwright version, matches the code | website-facts.test.js reads the Studio's source, website-commercial.test.js holds the prices to plans.ts | node --test tests/unit/website-facts.test.js tests/unit/website-commercial.test.js | 35 tests, 35 passed |
The examples on the three problem pages were run before they were written down: real bug or flaky, API schema drift and testing without a cloud grid. The output on each page is what the command printed, shortened where it says so.
Where do the dates on the pages come from?
The date in the sitemap of a page is the date of the last commit that touched that page, read with git log -1 --format=%cs by scripts/sitemap-lastmod.mjs. It is never the day we generated the file, because a date that moves every day teaches search engines to ignore it. The changelog lists published releases and, apart, the dated changes since.
What do these checks not show?
- A passing test shows that the code does what we describe. It does not show what result a customer gets. We have no customer outcomes to report, and the site publishes no testimonial.
- We have not benchmarked Nexus Studio against a cloud grid, so the site makes no speed comparison. The timings on the problem pages come from one laptop.
- The failing test on the home page is a staged demo, and the page says so. An estimate, such as the saving on the infrastructure page, is marked as an estimate, and its sums are checked to add up.
- A check is only as good as what it asserts. A behavior no test covers can be wrong without any check turning red.
What if a number here is wrong?
The product is right and the page is wrong, so we fix the page. Write to support@nexusstudioqa.com with the page and the number. The Trust page lists what we state about security and privacy, and the Verify a download page shows how to check an installer.