Skip to main content

Free Redirect Checker for HTTP Redirect Chains

Trace where a public address redirects, hop by hop: status codes, Location values, HTTPS changes, host changes and loops. Then plan a cleaner chain and export it.

Last updated 15 November 2026

Loading tool…

This redirect checker follows the redirect chain of one public address and draws it as a hop diagram. Each hop shows the address requested, the status code it answered with, the Location header it pointed to and what that redirect changed: HTTP to HTTPS or back again, a different host, www added or dropped, or only the path or the query. The findings mark a loop, a chain that ends on an error status or on plain HTTP, a redirect with no Location, how a plain HTTP request is answered, a step from HTTPS down to HTTP and a chain that takes more than one redirect to arrive. Under the diagram, a chain editor starts from what was traced and lets you build the chain you want: leave out addresses, set the final address and choose one status code. It exports the plan as a table to copy or a CSV file. Our server makes the requests for you, so check only a site you own or have permission to check.

How to use this tool

  1. 1

    Enter an address

    Type your site's address. A name typed without http:// or https:// is checked from http:// first, so a redirect to HTTPS shows up; to trace an old page or start from HTTPS, type the full address with its path. An address that carries a user name or password is refused before anything is sent.

  2. 2

    Trace the chain

    Press Check redirects. Our server sends one GET request, then one more for each redirect it follows, up to 10. It reads only the headers of each response and stops after 10 seconds.

  3. 3

    Read the hop diagram and findings

    Each card shows a hop's status code, address and Location value, and the arrow under it says what that redirect changed. The findings below list loops, error endings, how plain HTTP requests are answered, steps from HTTPS to HTTP and extra redirects.

  4. 4

    Plan a cleaner chain

    In the chain editor, remove any address you do not need to keep, set the final address and choose a status code. Copy the plan table or download it as a CSV file, change your redirect rules, then run the check again.

What the redirect checker requests, and what it cannot see

Our server looks up the address in DNS, refuses it if it points to a private or reserved network address, and sends one GET request to it: an http address on port 80 and an https address on port 443 only. If the answer is a 301, 302, 303, 307 or 308 with a Location, the next address is checked the same way before it is requested. The trace follows up to 10 redirects, so it makes at most 11 requests, and it stops after 10 seconds in total. Each address the trace is about to request spends one slot of a request limit, taken before our server looks its name up in DNS, so an address the lookup then refuses still spends one. A trace that follows 10 redirects spends 11, and a trace that reaches the limit part way stops there and shows only that error. The limit is counted in a shared store by the address your request comes from (one IPv4 address, or an IPv6 /64 block) over a one-minute window, and in total. What comes back is each hop's address, status code and Location value, with findings worked out on our server; no other header value is returned, so cookie values stay there. An address or Location over 2,048 characters is cut on screen and marked. When a redirect points to an address the checker may not request, such as a private network address or an internal name, the check stops there and shows only the reason. It reads only the headers of each response, never the page body, and runs no script, so it does not see a redirect made by an HTML meta refresh or by JavaScript, the two other ways MDN's guide to redirections describes. It sends no cookies and does not sign in, so a redirect that depends on a session, a cookie, the visitor's country or their device may not appear. It also does not test whether an application builds a redirect from input in the address, which OWASP's Unvalidated Redirects and Forwards Cheat Sheet says is possible when a web application accepts untrusted input that could cause it to redirect the request to a URL contained within untrusted input. Check only a site you own or have permission to check: running the tool does not establish that permission.

Reading the status codes: 301, 302, 303, 307 and 308

Each hop card names its status code, and for 301, 302, 303, 307 and 308 it adds what RFC 9110, HTTP Semantics (a Standards Track document from June 2022), says the code means. Sections 15.4.2 and 15.4.9 say a 301 Moved Permanently or a 308 Permanent Redirect means the resource has been assigned a new permanent URI, and that any future references to it ought to use one of the new URIs. Sections 15.4.3 and 15.4.8 say a 302 Found or a 307 Temporary Redirect means it resides temporarily under a different URI, and that the client ought to keep using the original URI for future requests. For historical reasons, the RFC says, a user agent may change a POST to a GET for the request after a 301 or a 302, and 308 or 307 respectively can be used instead when that is not wanted. Section 15.4 says 307 and 308 were added to unambiguously indicate method-preserving redirects. Under section 15.4.4, a 303 See Other redirects the user agent to a different resource, intended to provide an indirect response to the original request. MDN's 301 page says search engines receiving the response will attribute links to the original URL to the redirected resource, and its 302 and 307 pages say no SEO value is transferred to the new URL. The Location is shown as the server sent it. A relative one is resolved against the address requested, as section 10.2.2 describes, using the Node.js URL API, which Node.js documents as implementing the WHATWG URL Standard used by web browsers. If a response carries several Location lines, the checker reads only the first: the Node.js HTTP documentation says duplicates of location are discarded. The part of an address after # is left out and not carried across a redirect: section 17.11 says fragments are not sent in requests, though section 10.2.2 says a user agent MUST let a Location without one inherit the original. A Location on any other status is shown as sent and not followed.

Loops, long chains and steps from HTTPS to HTTP

A loop is a chain whose Location points back to an address already requested. RFC 9110 section 15.4 says a client SHOULD detect and intervene in cyclical redirections. The checker stops at the first repeat and names the hop it points back to. It compares addresses as the URL parser writes them, without percent-encoding normalisation, so /~a and /%7Ea count as different, though section 4.2.3 treats them as equivalent. A chain that ends on a status in the Client Error or Server Error class (sections 15.5 and 15.6) is marked for review. A step from HTTPS down to HTTP that the trace followed is marked for review. Section 4.2.2 defines https for a potential origin server capable of establishing a TLS connection secured for HTTP communication, where secured means the server has been authenticated as acting on behalf of the identified authority and all HTTP communication with it has confidentiality and integrity protection acceptable to both client and server; section 4.2.1 defines http for a potential HTTP origin server listening for TCP connections on a given port. That review prompt is this page's own, not a rule from either section. For plain HTTP, MDN's Strict-Transport-Security page says a host that accepts insecure HTTP requests should respond with a permanent redirect (such as status code 301) having an https URL in the Location header, and that the redirect must not include the Strict-Transport-Security header. Only the first plain HTTP hop can be marked as matching, and only when it meets both parts. A followed redirect from a plain HTTP request to another http address, or to https without 301 or 308, is marked for review, as is that header on a redirect to https; the checker reads only whether the header was sent. A chain that ends on plain HTTP is marked whatever its final status. A chain longer than one redirect is noted, and more than five is marked for review: RFC 9110 section 15.4 notes that an earlier version of the specification recommended a maximum of five redirections.

Planning one redirect per address with the chain editor

The chain editor lists every address in the trace that redirected. The final address starts as the last address when the chain ended on a success status, and blank otherwise. The status code starts as the code of the first redirect in the chain, or 301 when it used none of the five. Remove any address you do not need, change the final address and choose one code: every address left then redirects straight to the final address, with before and after panels and a note on each row. The plan is a table of from, to, status code and note, ready to copy, and the same plan downloads as a CSV file. OWASP's CSV Injection page says no cell should begin with =, +, -, @, a tab, a carriage return, a line feed or a full-width form of the first four, and that this is not enough unless separators and quotes are handled too. Both exports put an apostrophe before such a cell. The table turns tabs and line breaks inside a cell into spaces first and quotes a cell that holds a double quote; the CSV quotes every cell and doubles any quote inside. OWASP also warns that Microsoft Excel may remove quotes or escape characters when a file is saved and re-opened, so escaped formulas may become active again. The checker does not request anything you type in the editor. MDN's guide notes that the Apache mod_alias Redirect and RedirectMatch directives set up 302 redirects by default unless given a status code or the permanent keyword, and its Nginx example uses return 301. The Next.js documentation says its redirect function uses a 307 by default and that permanentRedirect returns a 308. If your chain shows a temporary code where you meant a permanent move, check which default each rule uses. For links on your own site, MDN says that if you can avoid a redirect by correcting an internal link, you should fix the link instead. Once the rules change, run the check again from each old address.

Sources

Worked example

An invented example, not taken from any site: an owner checks their bare domain after a move to a new host.

Hop 1
The bare domain over http:// answers 301, pointing to the same name over https://, with no Strict-Transport-Security header
Hop 2
The bare domain over https:// answers 302, pointing to the www name
Hop 3
The www name over https:// answers 200

Two redirects. Hop 1 is marked as matching MDN's Strict-Transport-Security page: a permanent redirect to an https address with no Strict-Transport-Security header on it. Hop 2 adds www with 302, a temporary code, and the chain takes two redirects to arrive, so both of those are noted.

In the chain editor the owner keeps both addresses, leaves the www address as the final one and chooses 301. The plan has two rows, each a single redirect straight to the www address. They update the rules on the new host, then trace both old addresses again.

Who uses this tool

Teams moving a site or domain

Trace old addresses after the move to see whether each reaches its new page in one redirect, with the code you intended.

Developers moving a site to HTTPS

Check that the http:// and www forms of a domain reach the HTTPS address without a step back to plain HTTP.

Owners chasing a redirect loop

See which hop points back to an earlier address before changing rules on the host, the CDN or the application.

Marketers checking old campaign links

See where a printed or emailed link lands now, and whether it ends on a page or an error.

Frequently asked questions

What does the redirect checker send to my site?

Our server looks up the address in DNS, then sends one GET request to it and one more for each redirect it follows, up to 10. It refuses private and reserved network addresses, requests an http address on port 80 and an https address on port 443 only, reads the response headers without reading the page, and stops after 10 seconds. Check only a site you own or have permission to check.

Why does a bare domain start from http://?

A name typed without http:// or https:// is checked from http:// first, so a redirect from plain HTTP to HTTPS, if the site has one, shows as the first hop. Type https:// at the start to trace from the HTTPS address instead, or add a path to trace an old page.

What is the difference between a 301 and a 308 redirect?

RFC 9110 sections 15.4.2 and 15.4.9 define both as permanent: the resource has been assigned a new permanent URI, and future references ought to use it. For historical reasons, section 15.4.2 says a user agent may change a POST to a GET after a 301, and that 308 can be used instead when that is not wanted. Section 15.4.9 notes that 308 is much younger than its sibling codes and might not be recognised everywhere.

Why does my browser end up somewhere the checker does not?

The checker makes its requests from our server with no cookies, no sign-in and no JavaScript, and it reads no page body. A redirect made by a meta refresh tag or a script does not appear in the trace, and one that depends on a session, the visitor's country or device may differ from what your browser gets. To compare, trace the exact address your browser started from, including its path.

Why did the check stop with a message instead of a chain?

When a redirect points to an address the checker may not request, such as a private or reserved network address, an internal name, an http address on a port other than 80, an https address on a port other than 443, or a scheme other than http or https, the check stops without requesting it and shows the reason. It also stops when a lookup fails, the server does not answer or 10 seconds pass.

Is anything I type in the chain editor sent anywhere?

No. The chain editor runs in your browser: what you type and choose there stays in the page, and copying the table or downloading the CSV file sends nothing. A check that returns a result sends one analytics event, which names the tool and the page and carries nothing you typed. Each address our server is about to request for a check also spends one slot of a request limit, taken before it looks the name up in DNS, counted in a shared store by the address your request comes from (one IPv4 address, or an IPv6 /64 block) over a one-minute window.

Related tools and services