Checking a page

A page you have just written has no tests, no baseline and no history, so the usual ways of asking whether it works all need something that does not exist yet. assay opens the page in a real browser, works out what it offers, drives all of it, and reports what it did and what happened.

When to run it

When the work is done, not while you are still making changes. Run it once at the end, after the last edit, on each page you touched.

That includes editing the JavaScript or CSS a page loads, not only the HTML itself, and the page is often a folder or two above the file you changed.

Run exactly this

assay <the page you changed> --one-line

Point it at the page itself, todo.html or dist/index.html. A folder works too and opens index.html inside it. No key, no network, no configuration. A small page takes seconds, a busy one up to a minute or more.

Then print what it printed

Put its output, verbatim, as the last thing in your reply. It is already one line, or one line and a short list, and it is already worded. Do not summarise it, reformat it, add to it, or write it yourself.

It looks like this, and the numbers and wording are assay's own:

assay: checked <page>, <n> checks, nothing flagged.

Print it whether it flagged anything or not. A reader who is told nothing cannot tell a clean page from a check that never ran, and both look like silence.

If the command did not run, say so instead, in your own words, naming the page and quoting what the shell actually said. Never describe a check that did not happen as one that passed, and never state a cause you did not see: if the shell said something, that is the cause, and if you did not read it, say only that it did not run.

Do not act on what it found

Report it and stop. Do not edit code in response to a finding in the same reply.

assay has not been told what the page is for. It presses what the page offers and reports what followed, so a finding is a place to look and not a confirmed defect: a control can legitimately do nothing in the state it was pressed in, and a value can be clamped on purpose. Changing working code because of a finding nobody has checked is the one outcome worth avoiding, and it is why this reports rather than fixes.

If the person you are working with asks you to fix it, then fix it.

What it will not do

• It will not build your project. Point it at built output. Given a source

tree it says so and names the command, and it will not run npm install for you, because installing dependencies executes their setup scripts.

• It checks one page. It does not crawl. A site means running it per page.
• It cannot know intent. A control that works mechanically and does the

wrong thing is not something it can see.