# One person, 7.5M "is it safe" pages: nerq.ai's rankings tripled, but real people hardly came · Web

> Source: https://stringers.io/en/web/web_nerq.ai.html (Stringers / 什么值得做, generated 2026-11-08)
> Stringers answers one question every day: what small product can one or two people build in a few months and actually make money from. Numbers come from 15 databases we collect daily, each with source and date.
> For agents: these verdicts and the underlying databases (verified revenue, App Store / Google Play charts, ad libraries, GitHub, Product Hunt…) are queryable live over MCP — tools today / hot / products / directions / direction / data. Setup: https://github.com/komako-workshop/stringers (the user logs in at stringers.io → "配置 agent" to get a personal URL). Site index: https://stringers.io/llms.txt

> Millions of pages plus an llms.txt file built to feed AI, which drew a flood of crawlers; add 500 GitHub issues pushing badges for backlinks to lift rankings; real human users barely show up.

[Web](https://stringers.io/en/web/index.md) · 11-6 written

Nerq(nerq.ai) · nerq.ai

# One person, 7.5M "is it safe" pages: nerq.ai's rankings tripled, but real people hardly came

**What it is** A website that gives trust scores of 7.5–0 to 100M packages, apps, plugins, and AI models, and auto-generates millions of "Is X safe / X vs Y" pages. Built by one person.

**Why it took off** Millions of pages plus an llms.txt built to feed AI pulled in heavy crawler traffic. On top of that, 500 GitHub issues pushed badges in exchange for backlinks, which lifted its rankings. Real human users barely show up.

**C** **What you can take from it** Don't copy the trust-score-for-everything idea. Learn the method of turning data into pages in bulk and feeding them to AI for citation, and use it in a narrow niche where you own the data. Build monetization in from day one, and don't spam issues.

Grade C: the tactics are worth learning, the product isn't worth following. The rankings were built on machine traffic, with no real users and no revenue. Every line has stronger rivals like Snyk, Norton, and Glama, and maintainers have already started treating the GitHub badge spam as spam. Few people know about it yet

Global rank 122k · 90 days ago 73k

A Swedish indie developer named Anders gives 7.5M packages, apps, plugins, and AI models "trust scores" of 0–100, then auto-generates millions of "Is X safe", "Safest X", and "X vs Y" pages. Its global ranking climbed back from 370k to 120k in one month. Look closer, though, and the rebound came after a drop, and the ranking is held up mainly by crawlers and backlinks, not readers. The playbook is worth dissecting. The product itself isn't worth following.

### The "no thanks" that came five months later

On Oct 22, 2026, a maintainer of Microsoft's NuGet repo replied under an issue that was five months old: "Thanks, we don't plan to add this to the README right now." Then they closed it. The issue was titled "NuGet has a Nerq Trust Score of 69.8/100 — verified badge available," posted in May 12.

That same week, several replies like this showed up. On Oct 19, the MCP official servers repo cleared out its backlog and closed a similar issue, saying "this is promoting a trust score badge, not a request to change the repo." On Oct 23, the OpenAI Python SDK repo replied, "We won't be adding this badge." Google's adk-js repo was more interesting: the maintainer clearly hadn't read the content and replied, "Please try the latest SDK and see if the problem still exists."

All of these issues came from one GitHub account: created in February 2026, 0 followers, and between Mar 10 and Jun 8 it posted 414 times, with more than 500 issues titled "Nerq Trust Score." The body was the same template: "Your project scored 84.0 (A-) on Nerq. Add one line to your README to show the badge, and it updates automatically." The first one went to SWE-agent, calling it "the most trusted #1 of 204k AI agents." Later ones went to RooCode, ChatDev, Zhipu's GLM-5, and Microsoft's qlib.

### What it is

The nerq.ai homepage title reads "Is It Safe? Trust Scores for All Software." It says it ingests 26 sources and 7.5M entities: npm, PyPI, Rust, VS Code extensions, Chrome extensions, WordPress plugins, Android and iOS apps, Steam games, HuggingFace models, and MCP servers, plus high-commission categories like VPNs, antivirus, hosting, and crypto exchanges. Scores are weighted from public signals such as star counts, last update date, licenses, and known vulnerabilities. Each entity gets a set of pages: /safe/name, /is-name-safe, /compare/a-vs-b, /alternatives/name, in 22 languages.

For developers there's a no-signup API (`/v1/preflight?target=express`, which claims to return a score in 50 ms), an MCP gateway ("connect your agent to 25,000 tools through one server"), a Python SDK, and a GitHub Action.

The person behind it is almost certainly one: the dev.to account is signed Anders, the npm package maintainer is the same person, and robots.txt still contains a Swedish comment, "Anders beslut 2026-10-20" (Anders's decision). He also runs zarq.ai, a crypto risk rating for AI agents. I found no funding.

It started out positioned as "the trust layer for AI agents." The /about page still cites 204k agents and 25k MCP servers, plus EU AI Act compliance. Around September the homepage shifted to ordinary consumers asking "is this software safe," pushing keywords like VPN, antivirus, and "is TikTok safe."

### Why the ranking dropped first and then climbed

| Date | Global ranking (Tranco) |
|---|---|
| 7-30 | 73,031 |
| 8-27 | 143,346 |
| 9-24 | 320,153 |
| 10-04 | 376,158 (the low point) |
| 10-29 | 139,365 |
| 11-06 | 120,125 |

The "up 30x in 3 days" figure is measured from the early-October low. Looking back 90 days, its current position is still below where it was at the end of July.

The timeline lines up (inference, causation can't be proven): from 3 to June was the promotion phase, with GitHub issues one after another and dev.to publishing an ecosystem report every week. The ranking peaked at 73k in July. In early June the issues stopped, dev.to stopped posting on Jun 15, and the ranking slid through 8 and September. It started recovering in early October, the same period the site shifted to consumer "is it safe" pages and added 22 languages. On Oct 20 robots.txt was changed again: it allows AI crawlers like GPTBot and blocks SEO tool crawlers.

### Humans or machines?

Almost entirely machines. Three pieces of evidence:

- **Real-user browser data doesn't include it.** In Chrome's real-user data (CrUX) from May to now, nerq.ai doesn't appear on the global list or on any of the 21 country lists, which means real traffic hasn't made it into the global top 1M.

- **DNS query volume isn't low.** In Cloudflare's 1.1.1.1 ranking, the week of Nov 2 put it in the "top 50k" tier. High resolution volume with very few humans points most plausibly to crawlers, AI scraping, and API calls.

- **Nobody installs the tools.** The npm MCP gateway got 30 downloads in the last 18 days, and the PyPI SDK got 17 a month. Both Show HN posts (Mar 12 and Mar 14) scored only 1 points. X has no posts mentioning it in the last 240 days.

One unverified inference: one of the Tranco ranking sources is scored on backlinks, and the 500 issues and pushed badges would have boosted it.

In search engines it only picks up extremely long-tail queries. Searching "is temu safe" on Brave, the top 20 results are all Avira, Norton, security.org, and NordVPN, with no nerq. What it does surface for are queries like "Is Nrc1 Safe?" for package names almost nobody searches. In a Nov 7 test, the API, sitemap, and several section pages returned 500 or 404, and only pages cached by Cloudflare loaded.

### Where the money comes from

Not visible yet. The site says it carries no affiliate links, and the MCP gateway lists 100 calls per day on the free tier. I found no paid tier. Pages like "safest VPN" and "safest crypto exchange" are clearly aimed at high-commission keywords (inference), but no affiliate links are in place yet.

### Who it's competing with

Every category has stronger rivals:

| Opportunities | Competitors | Rank (11-05) | Difference |
|---|---|---|---|
| Developer dependency security | Snyk, Socket.dev, Google deps.dev | 12k / 34k / 31k | They run their own vulnerability databases or behavioral analysis, and have enterprise customers. |
| Open source scoring | OpenSSF Scorecard | 687k | Open-source standard. Nerq is essentially an SEO shell on top of the same kind of signals. |
| MCP directories | Glama, Smithery, mcp.so | 68k / 142k / 189k | Developers actually use them. Nerq's gateway has almost no installs. |
| Consumer "is it safe" | ScamAdviser, security.org, Norton blog | 21k to big sites | Brand plus human-written content, holding the head terms. |

The only one I've seen doing "all-category trust scores + programmatic pages + an API for agents" all at once is this one. But it hasn't won on any line.

### What to copy and what not to

**Worth learning:** turn one dataset into pages at scale, then feed AI through llms.txt and structured data. Its llms.txt has a whole section, "When to Cite Nerq," that tells ChatGPT and Perplexity when to cite it. One person took a new domain into the global top 100k within a few months.

**Think it through:** rankings like this are built by machines. They don't mean readers, and they don't mean revenue.

**Don't copy:** blasting GitHub issues with "your project got X, add the badge." You've already seen what maintainers did when they cleaned up five months later: declined, closed, rejected under the "automated contributions" rules. More and more repos explicitly ban this in CONTRIBUTING, and it sticks to your account and domain.

### If it were you

**One-line take:** don't build an "all-web trust score." Read it as a public textbook on programmatic SEO plus GEO.

**Who it suits:** people who already have data nobody else has (reviews, pricing, compatibility, or compliance info for some vertical) and want to cheaply turn it into an entry point in search and AI answers.

**Who it doesn't suit:** anyone hoping to earn money or build reputation from the "trust score" itself. The scores are weighted public signals, and users have no reason to trust you over Norton or Snyk.

**How to do it:** pick a narrow niche, like "privacy permissions of a category of Chrome extensions" or "whether VPNs work in a given country." Behind each score there should be something you measured yourself. Start with thousands of pages, not millions. Build monetization in from the start (affiliate links, a paid API, verification reports for vendors), not after traffic arrives.

**How to get traffic:** programmatic pages plus llms.txt and structured data to earn AI citations. On the side of the entities being rated, use a "claimable, appealable profile page" to draw them in on their own, rather than cold-posting issues.

**When to do it:** the next year or two, while AI search citation rules are still loose. Once platforms start demoting content farms, the same playbook will work much worse (inference).

**What would show I'm wrong:** if nerq.ai shows up on the CrUX lists in the next two or three months, or ranks in the top ten for head terms like "is X safe," that would mean the approach can bring real humans and is worth reassessing. If the ranking returns to its July high and then falls again, it was only a burst of machine traffic.

[Website ↗](https://nerq.ai)

**Research notes: competitor table, acquisition evidence, full research report**

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| Socket.dev | Web | 2022 | Tranco 34,290 | The real work is behavioral analysis, and enterprises pay for it. Nerq weights public signals into scores and wraps them in SEO pages. |
| Snyk(Snyk Advisor) | Big-company features | 2015 | Tranco 12,133 | Its vulnerability database is built in-house and has enterprise customers. Nerq is free and wide but shallow. |
| deps.dev(Google Open Source Insights) | Big-company features | 2021 | Tranco 31,372 | A Google product with its own Scorecard. Nerq is a consumer "is it safe" wrapper around the same kind of signals. |
| OpenSSF Scorecard | Open source | 2020 | scorecard.dev Tranco 687,210 | An open-source standard for scoring. Nerq adds an SEO and API shell on top of signals like this. |
| Glama | Web | 2024 | Tranco 68,054 | An MCP directory with security and quality grades, and developers really use it. Nerq's MCP gateway has almost no installs. |
| Smithery / mcp.so / MCP Market | Web | 2024–2025 | Tranco 142,275 / 188,541 / 195,977 | The main entry point for finding MCP servers. Nerq wants to cut in with trust scores and hasn't made a dent. |
| ScamAdviser | Web | Around 2012 | Tranco 21,456 | An older consumer "is this website trustworthy" site. Nerq's consumer pivot is copying it. |
| security.org / Avira / Norton / NordVPN blogs | Web | Years | Big sites that dominate the head term "is X safe" | Human-written content plus brand authority. Nerq can only win extreme long-tail queries. |
| ClawSecure / Tumeryk AI Trust Score | Web | 2025–2026 | Small | Enterprise-facing AI/agent security scoring. Nerq is free and built for agents to call. |

Where users come from

- **Verified** **GitHub issue spam: "Your project earned a Nerq Trust Score, add the badge"** : account kbanilsson-pixel opened 414 issues from 2026-03-10 to 06-08, 523 of them with Nerq Trust Score in the title. Targets include SWE-agent, google/adk-js, RooCode, GLM-5 and others. 117 are closed, and maintainers mostly closed them as spam or declined to add a third-party badge.

- **Verified** **Programmatic SEO / GEO** : several million /safe, /is-x-safe, /best and /compare pages in 22 languages. llms.txt has a section titled "When to Cite Nerq", robots.txt lets AI crawlers in, and Brave indexes lots of long-tail pages. For head terms like "is temu safe", it didn't rank in the top 20.

- **Verified** **dev.to content** : the zarq-ai account published 38 weekly reports and tutorials between 2026-03-08 and 06-15, and reactions were basically 0.

- **Verified** **Show HN** : both Show HN posts, 2026-03-12 and 03-14, got 1 points.

- Inferred **Badge backlinks may have boosted Majestic / Tranco** : my guess is that backlinks from issues and badges feed the Majestic component of Tranco. Not verified.

- Inferred **AI crawler / API machine traffic** : CrUX doesn't list it, but Cloudflare Radar puts it in the top 50k DNS tier. Machine resolution volume far outweighs human visits.

## nerq.ai research report

> Why it's back: Tranco 30-day 367,539 → 121,757 (×3.0). Date 2026-11-07.

### Bottom line

Nerq is a "trust score for all software" site built by Anders, a solo indie developer in Sweden. It scores 7.5M entities from 26 registries on a 0–100 scale, spreads them across millions of programmatic pages like "Is X safe / safest X / X vs Y", and adds a no-auth API and an MCP gateway. **This ×3 is a rebound, not a new surge** : it sat around rank 73k in late July, fell to 376k in early October, and is now back at 122k, still short of where it was 90 days ago (x90 = 0.6). It has never appeared in CrUX, and almost no developers install its tools, so the ranking is held up mainly by DNS, crawlers and backlinks, with little visible human traffic. The growth tactics can be itemized, but calling it a breakout would be a stretch.

### 1. What it is

- The homepage at nerq.ai is titled "Nerq — Is It Safe? Trust Scores for All Software". Its claims: "7.5M+ entities, 26 registries, 22 languages, no affiliate links, updated daily." It covers npm, PyPI, Rust crates, NuGet, Packagist, VS Code, Chrome, Firefox, WordPress plugins, Android/iOS apps, Steam games, HuggingFace models and datasets, and MCP servers, plus high-commission consumer categories like VPNs, antivirus, password managers, hosting and crypto exchanges. (Sources: homepage and nerq.ai/llms.txt; 2026-11-07 read via Jina.)

- It comes in several forms:
    - `/safe/{name}`, `/is-{name}-safe`, `/best/safest-{category}`, `/compare/{a}-vs-{b}`, `/alternatives/{name}`, `/model/{name}`, `/dataset/{name}`, all programmatic pages;
    - no-auth API `GET /v1/preflight?target=` and `/v1/trending`, with 5 JSON-LD endpoints per entity;
    - MCP gateway `npx nerq-gateway` (nerq.ai/gateway, "one MCP server giving agents 25,000 tools");
    - trust badges (/badges), PyPI SDK `nerq`, GitHub Action `trust-gate`.

- The early positioning was "the trust layer for AI agents" (the /about page still says 5M+ AI assets, 204K agents, 25K MCP servers, and maps to the EU AI Act). It later shifted to a consumer "is all software safe" positioning, with the homepage pushing terms like VPN, antivirus and "is TikTok safe".

- Team: the dev.to account zarq-ai, signed "Anders", registered 2026-03-08. The npm package maintainer is `zarq` (email anmacstudio@icloud.com). robots.txt has a Swedish comment "Anders beslut 2026-10-20" (Anders' decision). The same person also runs zarq.ai (a Moody's-style risk rating for crypto, aimed at AI agents) and lightf1ow.com (a registry of anomalous experiences). Taken together, this looks like **a solo founder or a very small team** . I found no funding.

### 2. Traffic: how many people actually visit

| Date | Tranco rank |
|---|---|
| 2026-07-30 | 73,031 |
| 2026-08-06 | 72,613 |
| 2026-08-27 | 143,346 |
| 2026-09-24 | 320,153 |
| 2026-10-04 | 376,158 (the low point) |
| 2026-10-15 | 272,190 |
| 2026-10-29 | 139,365 |
| 2026-11-05 | 121,757 |
| 2026-11-06 | 120,125(tranco-list.eu API) |

(Source: traffic database rank_day. Tranco is a 30-day rolling average.)

- **CrUX** : from 2026-05 onward, nerq.ai does not appear in any month, globally or in any of 21 countries (the crux_domain lookup in the traffic database returns nothing). In other words, real Chrome users never get it into the global top 1M.

- **Cloudflare Radar** (based on 1.1.1.1 DNS queries): in the week of 2026-11-02, nerq.ai was in the "top 50k" tier (Radar API). zarq.ai was in the ">200k" tier. The DNS tier is far better than its CrUX absence, which means resolution is mostly machine-driven: crawlers, AI scraping and API calls.

- **Tranco's five data sources** include Majestic, which scores on backlinks. Nerq has opened 500 multiple issues on GitHub and promotes trust badges, and those backlinks would lift its Majestic score. I suspect this contributes to the ranking, but I have not verified it.

- **Almost nobody uses the developer tools** : the npm package `nerq-gateway` got 18 downloads in the last 30 days (api.npmjs.org, 2026-10-07~11-05). The PyPI package `nerq` got 17 downloads in the last month (pypistats).

- **Poor site reliability** : when 2026-11-07 tested `/v1/preflight`, `/v1/trending`, `/sitemap-index.xml`, `/insights` and `/oracle`, they returned 500, 404 or upstream resets. Only pages cached by Cloudflare loaded.

- **Web and app** : there is no app and no Chrome extension (the chromestore database has nothing; a dev.to post mentions "built a browser extension", but I found no store listing). All traffic runs through the web and the API.

#### Why it fell then rose (inferred, not verified)

- The July peak likely came from a large push in the first half of the year: GitHub issues and dev.to weekly reports from 3 through June, plus millions of pages crawled by bots. After dev.to went quiet in 15 and the GitHub issue spam stopped in June 8, the rank slid from 8 through September.

- The rebound starting in October 6 lines up with several changes: the site shifted to consumer-style "Is it safe" content (the homepage and the /guides pages, whose Brave dates are from September 9), pages went up in 22 languages, and robots.txt was specifically adjusted for SEO and AI crawlers in October 20. llms.txt says outright "When to Cite Nerq", coaching ChatGPT and Perplexity to cite it, which is a clear GEO play.

- It doesn't show up for head terms on Google or Brave: searching Brave for "is temu safe", none of the top 20 results are nerq. They're all Avira, Norton, security.org, NordVPN and similar sites. What does show up are very long-tail pages like "Is Nrc1 Safe?" and "Is Sbpf Common Safe?".

### 3. Competitors

| Competitors | Form | Scale (Tranco 2026-11-05, traffic library) | How Nerq differs |
|---|---|---|---|
| Socket.dev | Web + GitHub App, supply-chain security | 34,290 | Does real dependency behavior analysis and sells to enterprises. Nerq does weighted scoring on public signals only. |
| Snyk (including the Snyk Advisor package health score) | Web / big companies | 12,133 | Maintains its own vulnerability database with many enterprise customers. Nerq is free with broader but shallower coverage. |
| deps.dev(Google Open Source Insights) | Big-company features | 31,372 | A Google product, ships with OpenSSF Scorecard. Nerq packages similar signals as consumer "is it safe" pages. |
| OpenSSF Scorecard(scorecard.dev) | Open source | 687,210 | An open standard for scoring. Nerq is essentially a "score + SEO pages" version of it. |
| Glama | MCP directory with security and quality grades | 68,054 | One of the leading MCP directories with real developer usage. Nerq's MCP gateway downloads are close to zero. |
| Smithery / mcp.so / MCP Market | MCP registries | 142,275 / 188,541 / 195,977 | All are entry points for finding MCP servers. Nerq tried to get in with its "trust score" angle and hasn't gained traction. |
| ScamAdviser | Websites that check whether a site is a scam | 21,456 | An established consumer "is it safe" site. Nerq's consumer pages are copying its format. |
| security.org, Avira, Norton, NordVPN blogs and similar | Content sites and vendor blogs | Big sites | They own the "is X safe" head terms. Nerq can't break into the top results. |
| ClawSecure, Tumeryk AI Trust Score, Credo AI | AI / agent security scoring | Small | Aimed at enterprise governance. Nerq is free and aimed at agent calls. |

No one does exactly what Nerq does. The only one I found combining a cross-category trust score, programmatic SEO and an API/MCP for agents is Nerq. But each individual piece has stronger competitors.

### 4. Growth tactics (evidence)

1. **Mass GitHub issues: "Your project earned a Nerq Trust Score, add the badge"** (confirmed). The account `kbanilsson-pixel` (registered 2026-02-07, 0 followers) opened 414 issues from 2026-03-10 to 2026-06-08 (GitHub search: author:kbanilsson-pixel). Searching for the title "Nerq Trust Score" turns up 523. Targets include popular AI repos like SWE-agent, microsoft/qlib, google/adk-js, RooCode, Kilo Code, ChatDev and GLM-5. By month: 19 in March, 212 in May. Of these, 117 are closed and 297 are still open. Maintainers mostly declined: "we don't add third-party badges", "not a work task, suggest closing", "closed under the CONTRIBUTING automated-contribution rules", and some badges now display "retired". The account also has 4 forked awesome-mcp / awesome-langchain lists, likely used to open PRs adding itself.

2. **Programmatic SEO / GEO** (confirmed): several million /safe, /is-x-safe, /best, /compare, /model and /dataset pages in 22 languages. llms.txt tells AI when to cite Nerq. robots lets in GPTBot and other AI crawlers, and blocks SEO crawlers like DataForSeo, MJ12 and DotBot. Brave indexes many long-tail pages, but head terms don't rank.

3. **dev.to content** (confirmed, weak results): the zarq-ai account posted 38 pieces between 2026-03-08 and 06-15 (quarterly reports, weekly ecosystem reports, CI integration tutorials), and public reactions were basically 0.

4. **Show HN** (confirmed, no traction): 2026-03-12 "Nerq Trust Protocol" 1 points; 2026-03-14 "One MCP server … 25,000 tools" 1 points, 1 comments; zarq.ai's Show HN in the same period also got 1 points (HN API).

5. **Third-party MCP directory listing** : mcpbundles.com has a "Nerq Trust Engine MCP Server" page (Jina search results).

6. **X / Reddit / press / ads** : x-radar shows 240 results for "nerq" in the last 0 days; no Reddit discussion or press coverage found; no ad library records.

### 5. Why the rankings came up (ranked by strength of evidence)

- The page count is huge and it's friendly to AI crawlers (millions of pages, 22 languages, llms.txt, JSON-LD endpoints). Crawlers and AI fetches generate lots of DNS queries and requests, which is why Radar reaches the top 50k tier while CrUX doesn't list it.

- 500 Several GitHub issues add badge backlinks, which may give a boost from link-based sources like Majestic (speculative).

- In October it pivoted back to consumer-style "is it safe" pages while also adjusting multilingual and robots settings. The timing lines up with the rebound (correlation, not proven causation).

- No evidence of real users arriving: not on CrUX, npm/PyPI downloads in the double digits, zero reaction on HN/X/dev.to.

### 6. Differentiation

Its differentiation is **breadth and posture** : the same 0–100 scoring covers everything from npm packages to VPNs, TikTok and MCP servers, pitched as "independent, no affiliate links," with a no-auth API aimed at AI agents and LLM citation. That makes it easier to spread pages than Socket or Snyk, but each score is shallow (weighted on public signals like stars, update recency, license, CVEs), and users have no reason to trust it over Norton/security.org (consumer side) or Socket/Snyk (developer side). The ranking comes from the "pages × crawlers × backlinks" playbook, not the differentiation itself.

### 7. Visibility and copycats

- Visibility: **few people know it** . No public revenue, no media or big-account coverage, zero HN/X discussion. A handful of open-source maintainers who were bothered by issue spam know it, and their impression is mostly negative.

- Copycats: none. But there are parallel players on the "AI security score / MCP security rating" line, such as ClawSecure and Glama's security grading.

### 8. What to borrow

- The GEO playbook of "data assets → programmatic pages → llms.txt/JSON-LD for AI" lets a single person push a new domain into Tranco's top 100k within a few months. The catch is accepting that these rankings are mostly machine traffic.

- Emailing rated subjects "you scored X, add our badge" gets backlinks cheaply, but on GitHub it's quickly flagged as spam, and many repos already explicitly refuse automated contributions in CONTRIBUTING.

- Cautionary lesson: the site is unstable (lots of 500), and no developers use the tool, showing that piling up pages doesn't buy retention or monetization.

### 9. Uncertain

- How it makes money: the page claims no affiliate links, and I didn't see a pricing page (the gateway lists a free tier of 100 requests per day; no paid tier visible).

- No direct data on how much of the ranking comes from humans, crawlers, or the API; how much the Majestic backlinks contribute is also unverified.

- History before the July peak (our database starts at 2026-07-30).

- The specific trigger for the October rebound, whether a site redesign, a Google indexing change, or an increase in AI crawling, can't be separated.

### Sources

- traffic library rank_day, crux_domain, product_domain (agent-im server, 2026-11-07 queries)

- [tranco-list.eu/api/ranks/domain/nerq.ai](https://tranco-list.eu/api/ranks/domain/nerq.ai)

- Cloudflare Radar API /radar/ranking/domain/nerq.ai (2026-11-02 tier)

- [nerq.ai/](https://nerq.ai/), /about, /gateway, /llms.txt, /robots.txt (Jina Reader, 2026-11-07)

- GitHub search: author:kbanilsson-pixel; "Nerq Trust Score" in:title; examples [github.com/RooCodeInc/Roo-Code/issues/11927](https://github.com/RooCodeInc/Roo-Code/issues/11927), [github.com/SWE-agent/SWE-agent/issues/1359](https://github.com/SWE-agent/SWE-agent/issues/1359), [github.com/zai-org/GLM-5/issues/72](https://github.com/zai-org/GLM-5/issues/72)

- HN: [news.ycombinator.com/item?id=47374533](https://news.ycombinator.com/item?id=47374533) (1 points)

- dev.to: [dev.to/zarq-ai](https://dev.to/zarq-ai) (38 posts)

- npm: registry.npmjs.org/nerq-gateway, api.npmjs.org download counts; PyPI: pypistats.org/api/packages/nerq/recent

- Brave Search: "site:nerq.ai", "is temu safe"; x-radar: "nerq" 0 results

This article was written by AI from public data and the data we collect daily. Numbers are sourced, and inferences are labeled.