The cheapest fix on any technical audit
Broken internal links are unusual among SEO problems: they are entirely self-inflicted, entirely within your control, and unambiguous. There is no judgement call about whether a 404 is bad.
Each one wastes crawl budget on a dead URL, discards whatever authority the linking page was passing along, and drops a reader who was following your own navigation. They accumulate quietly - a slug edited here, a page retired there - and nobody notices because nobody clicks their own footer.
Empty anchor text, the quieter problem
A link with no anchor text passes no context about its destination. Usually it is an image or icon link with no accessible label, which means a screen reader announces "link" and nothing else. That is an accessibility failure first, and a wasted relevance signal second.
What this checks, and what it does not
One page, up to twenty unique internal links, HEAD first with a GET retry for servers that reject HEAD. Deliberately capped: this endpoint is open to anyone, and each check is a request we make to a host someone else named. An uncapped crawler behind a free form is a denial-of-service tool.
When the cap truncates, the result says so. A tool that quietly checks a subset and reports "no broken links" is worse than no tool.
Whole-site link graphs
Finding orphan pages, redirect chains and the pages nothing links to needs a full crawl with permission and time. That is what DraftCamp does once a site is connected - and it is where the interesting answers are, because the worst internal-linking problems are invisible from inside any single page.