Content Grader Test Harness

Twenty-four fixtures that keep the score honest

This page loads the Content Grader in a hidden frame and runs the scoring functions directly. The fixtures cover clean prose, spam, a listicle, a landing page, a short snippet, non-English text, HTML headings, a page with no H2s, numeric copy, a deliberately poor page, keyword targeting in and out of position, markup that breaks on mobile or for assistive tech, and link profiles with bare hosts, protocol-relative links, URL variants and fragments.

Waiting for the grader to load.

#	Fixture	Expected behaviour	Measured	Result
Run the fixtures to populate this table.

What is asserted

No non-numeric scores, no score above 100, a finite headline, a provisional short input, and the expected ordering between clean copy and spam or a deliberately poor page. The harness does not assert one exact score, because a small rule change should not fail every fixture.

What is not asserted

The harness does not call Google, an AI engine, or a paid API. It does not measure ranking, traffic, or citation. It tests the local scoring contract only.