Content Quality & Brain
The per-article scorecard showing how HarperFlow checked its own writing before you ever see it.
Every article has a Quality & trust tab showing an automatic self-check HarperFlow runs after writing it. It's meant to answer a simple question: how much of this can you trust without reading every word yourself? This article covers what's on it.
HarperFlow scores each article across six dimensions: factual grounding, citation quality, profile fidelity (how well it matches your brand), GEO optimization, writing quality, and formatting and semantics. Those roll up into an overall score and a verdict of pass, warn, or fail, along with a separate score for the article's generated image.
The panel shows how many sources the article cites and how many specific facts are grounded in them, so "well-researched" is something you can check rather than take on faith. You'll find this tab on the article's own detail view, right alongside its body, metadata, and citations.
Alongside the score, you'll see a short verification summary: how many numeric or factual claims were checked, how many were confirmed as accurate, how many were corrected to match their source, and how many couldn't be verified at all. Link verification covers this in more depth, along with how sources are checked before an article is even scheduled.
A warn or fail verdict doesn't hold an article back on its own. It runs after the article's already scheduled, and shows up as a Review suggested flag in your Articles queue so you know which upcoming articles are worth a closer read during your veto window. Whether to skip it is still your call.
Note: The audit runs in the background right after an article is created, so it never delays scheduling. A verdict can take a little while to show up on a brand-new article.