ToolBreezy.com - Broken Link Checker & Redirect Checker

Know what to fix, in what order, and which page to open. Every broken link and redirect, with the page that links to it.

Find the broken links on a page, and find out which page you have to go and edit. Every broken link checker tells you a URL is dead. Almost none tells you where it was linked from, which is the only thing you actually need in order to fix it. This one carries the referring page through every stage and puts it in the first column of the export. CHECK A SITE, NOT JUST A PAGE Every other tool of this kind checks the page in front of you. That is the wrong unit: a broken link belongs to a site, and knowing which page links to it only matters once there is more than one page it could be. One switch follows your site and checks every page it can reach from where you are. A dead link that appears on twelve pages is one finding with twelve referrers, not twelve problems. There is a limit on how many pages one run reads, and if it stops early it tells you, because a checker that quietly gives up looks exactly like one that found nothing. REFUSED IS NOT BROKEN, AND IT ASKS TWICE A 403, a 401, a 429 or a Cloudflare challenge means the server declined us. It does not mean the link is dead. Reporting those as broken sends you off to check a page that turns out to be perfectly fine, and after that you stop believing the tool. They get their own state, and so do links that simply could not be reached. Better than that: nothing is reported as a problem until it has been asked twice. A checker starts with a cheap HEAD request, and plenty of sites answer that badly while serving the real page perfectly well. Some say forbidden. Some say not found. Ask once and a working page looks dead; ask properly and it is fine. A lot of what other checkers report is the checker's own doing. THE TWO THINGS A STATUS CODE CANNOT SEE On your own pages it reads the page, not just the status, and finds the two broken links every other checker calls healthy. A page that answers 200 and is an error page anyway. Content systems that have lost a page often serve their "not found" template with a success code, so nothing complains and visitors land on nothing. A link to page#section where that section was renamed away. The server never sees the part after the hash, so the status is always 200 and the reader always lands in the wrong place. Both are read from the page's own title and headings, never its body text, so an article about designing 404 pages is not itself reported as one. A deleted page that now redirects to the home page is called out too. It answers 200 and looks perfectly healthy in every tool. REDIRECTS ARE A FINDING, NOT A FAULT A 301 is not an error. A link that quietly moves to another host, an http link that bounces to https, or a URL that only differs by a trailing slash are all worth knowing about. Each one shows you exactly where it ends up, so you can decide whether to update it. ONE FACT, SAID ONCE Some big sites refuse every automated request, whoever makes it. Check a page full of their links in an ordinary tool and you get a hundred identical rows of "forbidden" with your two real problems buried underneath. Here a site that refuses everything folds into a single line that says so, and drops to the bottom, because it is the one finding you do not have to act on. The links that are perfectly fine fold away the same way, so the list opens on what needs doing and still has everything else one click below it. SHOWS YOU THE LINK, NOT JUST THE URL Click any finding and the page scrolls to that link and outlines it. Knowing a URL is broken and finding it on the page are two different jobs, and most tools only do the first. Every address in a result is a link you can open, including the page that links to it, which is the one you actually have to go and edit. SEE DOFOLLOW, NOFOLLOW AND SPONSORED ON THE PAGE One button marks up every link where it sits: green passes link equity, red is nofollow, amber is a paid link, purple is user-generated, with a legend in the corner. A slider sets how thick the marks are drawn, because a sparse article and a crowded navigation page do not want the same weight. RECHECK ONLY WHAT FAILED A server having a bad minute is not a broken link. One button re-runs just the failures instead of all four hundred links, so a transient error costs you a second rather than another full pass. IT WILL NOT HANG, AND IT WILL NOT JUST SAY "ERROR" Every request has a deadline, so one server that accepts a connection and then goes quiet cannot stall the whole run. A link that runs out of time says so, because a slow server and a dead address are not the same problem. A browser will not tell an extension why a request failed: a dead domain, an expired certificate and a closed port all look identical. So a single failure stays honestly vague, and where there is real evidence it is used. Several dead links on one address, with nothing from that address ever answering, is not several findings, it is one, and it says the domain may be gone. POLITE BY DESIGN, PER SITE Fire four hundred requests at once and the server rate-limits you, and then a checker reports its own flood as a page full of broken links. Capping the total is not enough, because most links on a page point at the same site. Every site gets its own allowance here, so one busy destination is never hammered, and a site that asks us to slow down gets exactly that rather than an accusation. Other sites carry on at full speed meanwhile. You can pause, resume or stop at any point. STAY ON THIS SITE IF YOU WANT TO One switch decides whether the check ever leaves the site you are on. Check every link, or only your own pages and nobody else's servers. It is your browser and your address, so it is your call, and the choice is remembered. THE RUN SURVIVES THE POPUP A popup closes the moment it loses focus, and a page with hundreds of links takes minutes. The check runs in the background. Close the popup, do something else, open it again, and the run is still going. Results belong to the tab they were collected on, so another tab never shows you the last page's findings. EXPORT Copy every result to the clipboard, or save it as a CSV file that Excel opens without a fight, with the referring page as the first column. Saving needs no extra permission and adds no warning at install. WHAT IT ASKS FOR Nothing at install beyond the tab you are on. The permission to reach other sites is requested the first time you press the check button, because checking a link means asking that site whether it exists. Nobody should have to grant that just to see what the tool does. Free, no account, no tracking, nothing sent anywhere except the links you asked it to check.