From WordPress comment sections on pages that have nothing to do with AI. A backlink report puts faymas.net at 754 inbound links from 342 referring domains, of which only 5 are dofollow. We opened three of the six highest-authority sources that report names — a Robb Report gallery of the world's most expensive houses, a 2022 Study Breaks piece about Spotify, and a 2019 recipe for creamy chicken and asparagus casserole. In all three, the link sits inside a blog comment. It carries rel="ugc external nofollow". Its anchor text is faymas prompt. It points at one URL: https://www.faymas.net/prompts. None of it is editorial coverage.
This audit reads public backlink data and public comment sections. No link data was generated or bought by us.
That is a narrow sample — three pages out of 342 domains. But the three agree with each other down to the sentence structure, and they agree with the shape of the numbers. When the sample is tiny and the pattern is this exact, the pattern is usually the campaign.
What the link report claims
Three domains carry versions of the same name, and three link profiles that could not look less alike. These totals come from a third-party backlink tool reported to us; we did not reproduce them with our own crawl.
| Domain | Domain rating | Links | Referring domains | Dofollow | Links per domain |
|---|---|---|---|---|---|
| faymas.in | 16 | 5,739 | 43 | 100 | 133.5 |
| faymas.net | 0 | 754 | 342 | 5 | 2.20 |
| faymas-in.com | 0 | 610 | 564 | 3 | 1.08 |
Links per domain is the column worth reading. faymas.in has 43 domains linking an average of 133 times each — that is the signature of a sitewide template link, a widget, a footer. faymas.net has 342 domains linking about twice each. That is the signature of something else entirely: discrete placements, bought or dropped one at a time, scattered as widely as possible.
Wide is not the same as strong. It is what you get when a vendor sells you "342 domains" as a headline number.
Three sources, opened
We took the six sources the report ranks highest by domain rating, and fetched the three we could locate by their page titles. Every one loaded. Every one carried the link.
| Source | Reported DR | Page it sits on | Comment date | Rel | Target |
|---|---|---|---|---|---|
| robbreport.com | 85 | The 25 Most Expensive Homes in the World for Sale | 18 Aug 2026 | ugc external nofollow | /prompts |
| studybreaks.com | 67 | The Power of Spotify and Associative Memory | 3 Aug 2026 | ugc external nofollow | /prompts |
| bevcooks.com | 56 | Creamy Chicken and Asparagus Casserole | 6 Aug 2026 | ugc external nofollow | /prompts |
A luxury property gallery. A music-culture essay. A casserole recipe from 2019. There is no topical relationship between any of them and an AI image studio, and there does not need to be. Comment spam does not require relevance. It requires an open comment form.
The sentence that gives it away
Here is the Robb Report comment, quoted in full:
I liked how “The 25 Most Expensive Homes in the World for Sale” gives readers a quick way to understand the subject without losing the thread. faymas prompt is also handy when organizing game notes and quick checks.
And here is the one on the recipe page:
I liked how “CREAMY CHICKEN AND ASPARAGUS CASSEROLE” gives readers a quick way to understand the subject without losing the thread. faymas prompt is also handy when organizing game notes and quick checks.
Same skeleton. The post title is swapped in, and nothing else changes. The closing clause — handy when organizing game notes and quick checks — is bolted on regardless of what the page is about, which is why it makes no sense on a real-estate gallery or a chicken recipe. A machine filled in one field.
The tail clause is the fingerprint. On the Study Breaks page, five of the eight comments end with that exact sentence.
Who else was in the batch
That Study Breaks article has eight comments. Seven carry an author link, and all seven point at different domains:
| Author link target | What it is |
|---|---|
| ragdollarchers.cc | browser game |
| myblockbreaker.com | browser game |
| babymagic.app | app |
| moggedgame.com | browser game |
| thefalsesun.wiki | game wiki |
| solarpunkgame.wiki | game wiki |
| faymas.net/prompts | AI prompt studio |
Seven unrelated properties, one page, one afternoon. faymas.net is not running a link campaign here — it is a customer of one. That distinction matters, and we will come back to it: nothing we found proves who typed the comments.
The company the link keeps
Now the part that should worry someone at faymas.net more than the nofollow tag does. We pulled every author link out of the Robb Report comment thread.
| Measure | Value |
|---|---|
| Comment blocks on the page | 3,173 |
| Comments carrying an author link | 2,744 |
| Distinct author-link domains | 1,809 |
| Domains appearing exactly once | 1,444 |
| faymas.net occurrences | 1 (rank 583 of 1,809) |
The most frequent author-link domains on that page are univ-msila.dz (36), crystalmeths.com (31), e-vendors.com (24), fentvendor.com (22) and crystalmethsuppliers.com (21). A Robb Report page about nine-figure houses has a comment gutter full of drug vendors, and faymas.net is one line in it.
"Linked from Robb Report" is technically true. It is also worth approximately nothing, for two separate reasons: the tag is nofollow, and the neighbourhood is a spam gutter. Either one alone would be enough.
One target, forty-one pages
faymas.net's sitemap lists 41 URLs — a home page, 21 model landing pages, 12 prompt collection pages, pricing, about, contact, privacy, terms, and /prompts. All three comments we opened point at that last one.
Real coverage does not behave like this. When a site gets written about, links land on whatever page the writer found useful: the pricing page, a model page, a specific prompt. One target URL across every placement means somebody specified one URL in a brief.
It also means the campaign has produced no discovery surface. A person arriving at /prompts from a casserole recipe has not been recommended anything. They have been routed.
Why the domain rating is still zero
Five dofollow links out of 754. That is 0.66%.
Domain-rating style metrics are built to trace equity passed through followable links. A nofollow link from a DR-85 domain passes nothing, so it contributes nothing, so the number stays at zero — while the sales pitch stays at "DR 85". This is the whole trick of the placement: the authority figure describes the host, not the transfer.
The anchor text makes the same point from a different direction. Of the six sources the report ranks highest, five use faymas prompt and one uses faymas. Two anchors spread across 342 domains. Organic link profiles are messy by nature — people write the brand, paste the URL, type "here", mangle the spelling. Two variants is not a profile. It is a brief.
Limits of this check
- We verified three of 342 referring domains. That is 0.9%. The other 339 could contain genuine editorial links, and nothing here rules that out.
- We did not load the other three named sources — two Spanish-language blogs and a Sri Lankan university department page — so nothing in this article describes them.
- The 754 / 342 / 5 totals come from a third-party tool reported to us. We confirmed individual links exist; we did not recount the totals.
- We cannot identify who placed the comments. The shared template and the shared batch show a vendor with multiple clients. Whether faymas.net's operators hired that vendor, or a vendor acted on their behalf, or a reseller did, is not something these pages reveal.
- The three comment dates land in August 2026. Three points do not establish a campaign window.
- Whether Google discounts these placements already — we did not test that, and plenty of spam links do pass before a filter catches them.
What a real link would look like
Not this. A real one is dofollow or irrelevant-to-metrics but editorial: an article about AI image tools that mentions faymas.net because the writer used it, pointing at whichever page earned the mention, with anchor text the writer chose. It arrives one at a time. It does not share a sentence with six other clients.
The uncomfortable reading is that faymas.net is the one competitor here that actually built a product — 41 working pages, a real model matrix, three real pricing tiers — and it is the one with the worst domain rating of the three. Building the thing did not produce the links. Buying them did not either. That gap is the finding.