Content Quality & Brain
The two checks HarperFlow runs on an article's sources and links before it ever reaches your schedule.
Citing real, working sources is part of what makes HarperFlow's articles trustworthy to both readers and AI search systems. Two separate checks make sure of it.
Before an article is scheduled, HarperFlow checks every source link it plans to cite to confirm it resolves to a real, reachable page, not a dead link or a search-results page that got swept up during research by mistake. A cap limits how many citation links a single article carries, and when good sources are thin, HarperFlow adds real links back in rather than shipping an under-sourced draft.
Separately, every article passes through a gate that checks the links themselves for problems your CMS shouldn't carry: leaked search-engine result links, a link whose visible text names your brand but points somewhere you don't own, or a link to a competitor's domain you've told HarperFlow about. Any of these gets automatically corrected or removed. A hallucinated wrong-domain version of your own site, your .com cited instead of your real .io, for instance, gets quietly rewritten to the domain you actually own, rather than stripped outright.
If that gate catches a real problem, or an article ends up with too few genuine cited sources, HarperFlow holds it in pending review instead of putting it on the schedule automatically. You'll see it as "not on the schedule yet" rather than with a publish time, and it needs a slot from you before it goes anywhere; see Article states.
Tip: If an article lands in pending review instead of scheduled, check its Quality & trust tab first. Thin sourcing is the most common reason, and the citation count there tells you immediately.