uj-ng.com — AI Site Grade

Analysis

UJNG's sophisticated AI-agent infrastructure (llms.txt, agents.md, UCP endpoint) is fully operational, yet the actual storefront is intermittently blocked behind Cloudflare challenges — creating a paradox where AI crawlers are welcomed in documentation but walled off from the content they need to read.

Crawler Access

Every AI bot tested — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended, ChatGPT-User, anthropic-ai — receives a 403 Cloudflare challenge on the homepage, all collection pages, all product pages, and the blog index. The robots.txt contains no AI-bot-specific rules (no Disallow for any AI crawler), meaning the blocking is entirely at the Cloudflare edge layer, not the robots.txt. The homepage and product pages were briefly accessible earlier in testing (returning 200 with full content), then flipped to 403 — indicating intermittent Cloudflare JS-challenge behavior that makes AI-crawl reliability unpredictable.

AI-Agent Infrastructure (The Paradox)

The site has an unusually advanced AI-agent stack for a Shopify store. An llms.txt exists at /llms.txt (4296 bytes), an agents.md at /agents.md, and a fully populated Universal Commerce Protocol (UCP) endpoint at /.well-known/ucp returning a 3860-byte JSON manifest with shopping, checkout, fulfillment, discount, and catalog services. The sitemap even includes a dedicated sitemap_agentic_discovery.xml. This infrastructure instructs AI shopping agents how to search, cart, and checkout — but none of these agents can read the product catalog because every product and collection URL returns a Cloudflare 403 challenge.

Cold-Knowledge Gap

A frontier LLM queried cold about "UJNG fashion retailer" returned: *"I do not have specific, verified information about a fashion retailer named UJNG... It is not a widely recognized brand in my training data."* The site itself sells luxury brands like Guidi, Ziggy Chen, Issey Miyake, Heliot Emil, UMA WANG, and Simone Rocha — designers with strong cult followings — but the LLM has zero prior knowledge of UJNG as a retailer. The site's homepage text (758 words) describes it as "globally sourced luxury & independent designers" with free international shipping, but no external signals (reviews, press, Reddit threads, search results) surfaced anywhere to corroborate the brand's existence or reputation.

Schema Posture

The homepage carries a single Organization schema with name, URL, contact phone, and social links (Instagram, Facebook). The Guidi product page has a well-formed Product schema with brand, offers (all marked OutOfStock), price in EUR, and SKUs. However, blog posts carry zero schema markup — no Article, BlogPosting, or FAQPage — despite containing substantive editorial content about designers. The blog index and the designers page are both Cloudflare-blocked, so their schema cannot be assessed.

External Signals

No external mentions of UJNG were found across web search, Reddit, or general indexing. The domain resolves to Shopify (IP 23.227.38.65) behind Cloudflare, with Google DNS and Google MX mail. The copyright notice reads "2026, UJNG Powered by Shopify" — a year ahead of the current date, suggesting a template default or a forward-dated deployment. The site has been on the Wayback Machine since at least December 2023, indicating it has existed for over a year with virtually no external footprint.

Findings

1. All AI crawlers blocked by Cloudflare JS challenges on storefront pages High

   Every AI bot tested receives a 403 Cloudflare challenge on the homepage, collection pages, product pages, and blog index. The blocking is intermittent and occurs at the Cloudflare edge layer, not via robots.txt.

   What to change: Configure Cloudflare to allow AI crawlers (e.g., GPTBot, ClaudeBot) to bypass JS challenges, or serve static HTML versions of key pages to bots.

2. AI-agent infrastructure invites agents but blocks access to catalog High

   The site has llms.txt, agents.md, and a UCP endpoint instructing AI shopping agents how to interact, but every product and collection URL returns a 403 challenge, making the infrastructure unusable.

   What to change: Ensure that pages referenced in llms.txt and UCP are accessible to AI crawlers without Cloudflare challenges.

3. No external mentions or backlinks found for UJNG High

   Web searches for UJNG, including reviews, Reddit threads, and general mentions, returned zero results. The brand has no external signals to corroborate its existence or reputation.

   What to change: Build external signals through PR, influencer collaborations, and backlink acquisition to improve brand recognition and AI visibility.

4. LLMs have no prior knowledge of UJNG as a retailer High

   A frontier LLM queried cold about UJNG returned no specific information, indicating the brand is not recognized in training data despite selling well-known designer brands.

   What to change: Increase brand visibility through content marketing, structured data, and external citations to improve LLM knowledge.

5. Blog posts carry no Article or BlogPosting schema markup Medium

   Substantive editorial content about designers lacks any structured data, reducing chances of appearing in rich results or being understood by AI crawlers.

   What to change: Add Article or BlogPosting schema to all blog posts with author, date, and headline properties.

6. Product schema marks all offers as OutOfStock Medium

   The Guidi product page schema lists all offers with OutOfStock availability, which may deter AI agents from recommending the product.

   What to change: Ensure product schema reflects actual stock status; consider removing out-of-stock variants or marking them as PreOrder if applicable.

7. Copyright notice shows year 2026, ahead of current date Low

   The footer reads '2026, UJNG Powered by Shopify', which may appear as a template error or inconsistency to users and crawlers.

   What to change: Update the copyright year to the current year or use dynamic year rendering.

8. Cloudflare JS challenges cause intermittent access for all crawlers High

   The homepage and product pages were accessible earlier but later returned 403, indicating unpredictable Cloudflare behavior that undermines reliable AI crawling.

   What to change: Disable JS challenge for known AI bot user agents or serve static cached versions to bots.

What's working

• llms.txt file provides AI crawlers with site guidance — A 4296-byte llms.txt file exists, offering structured guidance for AI crawlers about the site's content and capabilities.
• agents.md file provides agent-specific instructions — An agents.md file is available, indicating the site has considered AI agent interactions.
• Universal Commerce Protocol endpoint returns full manifest — The UCP endpoint at /.well-known/ucp returns a 3860-byte JSON manifest with shopping, checkout, fulfillment, discount, and catalog services.
• Sitemap includes dedicated agentic discovery sitemap — The sitemap references a sitemap_agentic_discovery.xml, showing proactive planning for AI agent navigation.
• Homepage has Organization schema with contact and social links — The homepage includes a valid Organization schema with name, URL, phone, and social media links, aiding entity recognition.
• Product page includes Product schema with brand and offers — The Guidi product page has a well-formed Product schema with brand, offers, price, and SKUs, though all marked out of stock.
• robots.txt does not disallow any AI bots — The robots.txt file contains no Disallow rules for AI crawlers, meaning blocking is not intentional at that level.
• Site has been indexed by Wayback Machine since 2023 — Wayback Machine snapshots exist from December 2023, indicating the site has been online for over a year.