Is shepsky-ai.pages.dev a scam? TrustSniffer's high-risk assessment: 5/100

Overview

This website is a project-style deployment under the name "shepsky-ai", published on a free shared-hosting subdomain of a developer deployment platform rather than on a standalone registered domain. When automatically visited, it presented not an application or storefront but an access-control interstitial, titled as a suspected-phishing warning, so its commercial purpose and business model cannot be established from its own content.

Scam/Impersonation Risk

Contradictory signals are present, and this section carries the weight of the assessment. Two independent reputation sources currently flag the site as malicious, and a further three treat it as suspicious. Together they amount to a confirmed blacklist hit. The same underlying signals associate the site with known phishing activity, so it is treated as a potential impersonation or credential-harvesting vector. No conflicting legal entities were found, because the site exposes no legal or identity pages that could be compared. The sandbox's behavioral telemetry showed no wallet-connection prompts, clipboard manipulation, hidden forms, external password forms or device fingerprinting. All external requests went to recognised, whitelisted destinations. That reduces the likelihood of an active drainer or script-injection payload during the observed sessions. It does not offset the external threat detections, because the site's real content was never rendered and phishing infrastructure often serves its payload conditionally. The site's own front page offers no brand claims, product descriptions or contact details to analyse, so the alarm rests entirely on external reputation and on the platform's own warning interstitial. Evidence:

• Homepage: the page captured on automated access was a warning interstitial titled as suspected phishing and returned a blocked (403) response. No site-authored content was captured.
• Reputation: two sources rate the site malicious and three rate it suspicious. A broader external blacklist flag is also set.
• Domain provenance: the domain is about 1.7 years old. Its first archive snapshot dates from 2024-11-28 and its latest from 2026-08-03. The archive holds 12 snapshots across a tracked span of 3 years, with continuous coverage and no repurposing detected.
• Behavior: across 3 successful sandbox runs there were 19 external XHR calls and 5 external POSTs, all to whitelisted destinations. The sandbox also recorded 15 console errors and an external-request share of up to 94.9% on the busiest run.

Regulatory Verification Notes

The site discloses no operator name, registered company, licence or registration number, or terms and privacy pages in the content that could be captured. Disclosure is therefore unclear and ownership is opaque. Opacity of this kind is an administrative and transparency limitation, not an identity contradiction in itself. It does matter here, because the site sits beneath a high-severity external reputation signal. Nothing links the site to a regulated entity, and no third-party broker or regulatory-database record matched it. The absence of that record is neutral and is not evidence of wrongdoing. The valid, well-formed transport-encryption certificate is a baseline hygiene signal only. It says nothing about who operates the site. Evidence:

• Legal and identity pages: no operator, company or licence disclosure was observed in the captured content.
• Hosting: served from AS13335 (Cloudflare, Inc., US), registry ARIN.
• Transport encryption: TLS certificate issued by WE1, domain-validated (DV) only, valid to 2026-12-08, with about 42 days remaining.
• Broker intelligence: no match in the third-party broker reputation dataset.

What to Verify Next

Any engagement should be preceded by independent confirmation of who operates this deployment. The first step is to establish through an offline or otherwise independent route whether any known organisation claims the "shepsky-ai" name. If one does, its official domain, which should be a standalone registered domain rather than a shared deployment subdomain, should be obtained directly from that organisation and compared. Anyone who reached the site through a link, message or advertisement should trace that referral to its sender and confirm the sender's authenticity through a separate channel. If any licensing or regulated activity is claimed, it should be checked against the relevant registry using the entity's own published details. Official support and contact channels should be confirmed independently before any communication. Evidence:

• Homepage: no brand, operator or regulatory claim is available to check, so verification has to start from the referral source.
• Hosting model: the site runs on a shared platform subdomain, which means the registered domain itself carries no standalone brand identity.
• Domain history: continuous but short archive history (first seen 2024) gives no long-term reputation to lean on.

Summary Verdict

The available evidence supports an elevated-risk posture. Confirmed adverse reputation signals and phishing association outweigh the site's technically orderly setup, and no verifiable operator identity exists to counterbalance them. The site is not suitable for any sensitive interaction in its current state.

Infrastructure Integrity

The site is fronted by Cloudflare's CDN and web-application firewall, and every observed address belongs to its edge network, so the origin server was not observable. That is a mitigating factor for availability and basic attack resistance, and the platform's bot-challenge layer behaved consistently during testing. It confers no legitimacy and cannot offset the external threat detections. Edge-only visibility also prevents independent verification of where the content is really served from.

Closing Assessment

The site should be treated as unsafe for login, credential entry, payments or any financial interaction. At most, it could be viewed passively, with no data entered, no files downloaded and no wallets connected. Anyone who has already submitted credentials or funds should change the affected passwords and contact their bank or payment provider promptly.