Not when they redirect; yes when they render. On 19 November 2026 we sent eight HTTP methods to faymas.in, faymas.net and faymas-in.com, repeating every cell three times. All 24 redirect cells gave the same status code and the same Location no matter which method we used. Behind them the applications disagreed: TRACE drew 405 on all three, OPTIONS drew 400 / 400 / 200, and one invented verb drew 400 / 501 / 400.
This guide analyses published prompt text. No generation tests were run.
What we sent, and what we refused to guess
We used the canonical host for each domain, as established by the redirect audit on 17 November: https://faymas.in/, https://www.faymas.net/ and https://faymas-in.com/. Eight methods went to each root: GET, HEAD, POST, PUT, PATCH, DELETE, OPTIONS, TRACE. Four of those methods then went to six URL variants: three double-slash paths, two non-canonical hosts and one plain-http URL. That is 48 cells, three repeats each, 144 responses. All 48 were stable across repeats. A follow-up set of 42 responses covered non-standard methods, full error bodies and the apex-versus-www double slash.
POST, PUT and PATCH carried a 22-byte form body with a recognisable token. No form endpoint was targeted. No cookie, login state or authentication header was sent. No redirect was followed, so every result below describes the first hop only. Header names are quoted with their original capitalisation because the exact spelling is a fingerprint of the sender, not part of HTTP semantics.
Before writing, we searched every one of the 51 live guide pages for the exact strings POST, 405, PUT, PATCH, DELETE, OPTIONS, TRACE, Method Not Allowed and preflight. Zero hits. The only near misses were two pages that explicitly left this hole open. The 17 November page says: "We did not measure POST behaviour, 308 method preservation in practice". The 18 November page says: "We did not send POST, and did not test how a 308 preserves a body". This page closes that hole. It does not redo either audit.
The redirects never looked at the method
Every redirect we tested ignored the verb. Four methods, six variants, 24 cells. Same status. Same Location. Same capitalisation of the header name. Three repeats each.
| URL variant | GET | POST | PUT | OPTIONS | Location, as sent |
|---|---|---|---|---|---|
https://faymas.in// | 308 | 308 | 308 | 308 | location: / |
https://www.faymas.net// | 308 | 308 | 308 | 308 | Location: / |
https://faymas-in.com// | 200 | 200 | 200 | 200 | none |
https://www.faymas.in/ | 301 | 301 | 301 | 301 | location: https://faymas.in/ |
https://faymas.net/ | 301 | 301 | 301 | 301 | Location: https://www.faymas.net/ |
http://faymas.in/ | 301 | 301 | 301 | 301 | Location: https://faymas.in/ |
One thing did change with the method, and it was not the redirect. On https://faymas.in// the GET response carried cf-cache-status: BYPASS while POST, PUT and OPTIONS carried DYNAMIC. On https://www.faymas.in/ the GET was a HIT with cache-control: max-age=300; the other three were DYNAMIC with no cache-control at all. The cache layer reads the verb. The redirect rule does not.
Three stacks, eight verbs
At the canonical root, six of eight methods returned 200 on all three domains. Two did not. The table gives the status codes; the byte counts are the HTML bodies behind them.
| Method | faymas.in | faymas.net | faymas-in.com |
|---|---|---|---|
| GET | 200 / 617,289 B | 200 / 314,758 B | 200 / 71,239 B |
| HEAD | 200 / 0 B | 200 / 0 B | 200 / 0 B |
| POST | 200 / 594,660 B | 200 / 314,810 B | 200 / 55,935 B |
| PUT | 200 / 603,425 B | 200 / 314,758 B | 200 / 55,935 B |
| PATCH | 200 / 589,736 B | 200 / 314,758 B | 200 / 55,935 B |
| DELETE | 200 / 585,461 B | 200 / 314,810 B | 200 / 55,935 B |
| OPTIONS | 400 / 0 B | 400 / 0 B | 200 / 55,935 B |
| TRACE | 405 / 155 B | 405 / 155 B | 405 / 155 B |
Every one of the 24 cells was stable across three repeats. A 200 here does not mean "the POST was accepted as a form submission". It means a page was returned. Our token never appeared in any response body, so we did not find evidence that the body was read back to us. Whether an application consumed it internally is not observable from a first-hop response.
The one refusal all three share
TRACE produced a byte-identical answer on all three domains. Not similar. Identical.
| Field | Observed value, all three domains |
|---|---|
| Status line | HTTP/1.1 405 Not Allowed |
| Server | cloudflare |
| Content-Length | 155 |
| CF-RAY | - |
| Body | <center><h1>405 Not Allowed</h1></center><hr><center>cloudflare</center> |
| Allow header | absent on all three |
| x-powered-by / via / cf-cache-status | absent on all three |
No application signature appears anywhere in that response. No Next.js, no PHP, no Caddy, no LiteSpeed. The three sites run three different stacks, yet this refusal is one page. That points to a rule applied at the edge, before the request reaches an application. We report the missing Allow header as an observation. We do not turn it into a compliance verdict.
One invented verb, three answers
A verb the server has never heard of turned out to be the sharpest probe. We sent PROPFIND and a made-up FOOBAR, three times each.
| Method | faymas.in | faymas.net | faymas-in.com |
|---|---|---|---|
| PROPFIND | 200 | 200 | 200 |
| FOOBAR | 400 | 501 | 400 |
PROPFIND is a real WebDAV method and all three returned a page. FOOBAR is not a method at all. Two domains answered 400. faymas.net answered 501 Not Implemented — the only 501 in the entire audit. That is a difference no URL variant exposed.
OPTIONS split the three as well. On faymas.in the 400 carries via: 1.1 Caddy, a vary: rsc, next-router-state-tree, ... list and cf-cache-status: DYNAMIC. On faymas.net the 400 carries x-opennext: 1 and the same RSC vary list, but no via. On faymas-in.com the 200 is a full 55,935-byte homepage with x-powered-by: PHP/8.5.12, platform: hostinger, x-turbo-charged-by: LiteSpeed, and this line: x-two-optimize-reason: Request mode is: OPTIONS. One application states outright that it noticed the verb. The other two reject it.
What "a 308 keeps the method" means on these domains
The 17 November audit wrote: "A 308 keeps the request method and body. A 301 does not promise that." That is a statement about the client's next request. A server issuing a 308 does not forward a body anywhere. It returns a status and a Location. Preserving the method and replaying the body is something the client does afterwards, if it obeys the status code. Our probe never followed a redirect, so the preservation itself is outside these measurements. What we can say is narrower and firmer: the decision to emit 308 or 301 did not depend on the verb in any of the 24 cells.
That earlier page also called the faymas.in double slash "the only 308 in the whole matrix". It was, in that matrix. The 17 November probe used apex variants; https://faymas.net// returned 301 to the www root. We re-ran that exact URL three times and got the same 301. We then used the canonical www host, https://www.faymas.net//, which was not in that matrix, and got 308 to a relative /. Both records are true. They are different URLs. Even the two 308s disagree on form: faymas.in sends lowercase location: / with via: 1.1 Caddy; faymas.net sends capital-L Location: / with no via. Capitalisation is not HTTP meaning. It is a fingerprint of who wrote the response.
A different method, a different number of bytes
Status codes were mostly the same. The bytes were not. That is where the method leaves a mark.
| Domain | GET | Non-GET | Cache headers |
|---|---|---|---|
| faymas.in | 617,289 B, HIT, max-age=300 | 585,461 – 603,425 B, all different from each other | DYNAMIC, no-cache, no-store, max-age=0 |
| faymas.net | 314,758 B | 314,758 or 314,810 B only | no-cache, no-store, max-age=0, no cf-cache-status |
| faymas-in.com | 71,239 B | 55,935 B for every non-GET method | DYNAMIC, PHP/8.5.12 |
On faymas-in.com the drop is 15,304 bytes, about 21.5 percent. On faymas.in no two non-GET bodies matched. On faymas.net the body stayed inside a 52-byte band. Three stacks, three ways of rendering a verb they were not expecting. None of this is visible if you only record the status code.
Limits
- We did not follow redirects. Method and body preservation on the second hop was not measured.
- We did not submit to any form endpoint, send cookies, or test a logged-in state.
- The token in the POST body was not echoed in any response. We did not find a reflection; we cannot say what happened inside the application.
- Only root paths and six URL variants were probed. Only the canonical hosts were used for the method matrix, plus three apex or www comparisons.
- QUIC and UDP 443, other network vantage points, browser cache behaviour after a redirect, TLS session reuse, DNSSEC and certificate transparency logs were not tested.
curl -Xconstructs a request line; it is not a browser. A real browser would not normally send PUT or FOOBAR to a page.- We did not find registrant, origin server or account ownership data. No traffic, ranking, revenue or search-volume figures appear on this page.
Where a result is a tool limit rather than a property of the site, we say so instead of turning it into a finding.
A URL tells you where a site sends you. A verb tells you which piece of the stack answered before any page was built. The redirect audit already gave us the first. The second cost one invented word and three repeat requests. Neither view is a popularity signal, and neither identifies who owns these domains. They are simply two cheap ways to separate an edge rule from an application, on three sites that share a word and very little else.
FAQ
Do the Faymas redirects change when you send a POST instead of a GET?
No. All 24 redirect cells across six URL variants returned the same status code and the same Location for GET, POST, PUT and OPTIONS, three repeats each.
Which HTTP method do all three Faymas domains refuse?
TRACE. All three returned 405 Not Allowed with the same 155-byte Cloudflare page and CF-RAY: -, and none sent an Allow header.
Does a 308 redirect preserve the request body on these sites?
We did not measure that. A server issuing a 308 sends a status and a Location; preserving the method and body is the client's next request, and our probe did not follow redirects.
Do the three Faymas domains agree on OPTIONS?
No. faymas.in and faymas.net returned 400; faymas-in.com returned 200 with a 55,935-byte page and a header reading "Request mode is: OPTIONS".
Is the double slash on faymas.net always a 301?
No. On the apex it is a 301 to the www host; on the canonical www host it is a 308 to a relative /. The 17 November page used the apex variant, so both records can be true.