Skip to main content

Command Palette

Search for a command to run...

Threat Analysis: Investigating Suspicious Outbound Connections to cdn.jsdelivrs[.]com (Adware or Supply Chain Foothold?)

Updated
3 min read
Threat Analysis: Investigating Suspicious Outbound Connections to cdn.jsdelivrs[.]com (Adware or Supply Chain Foothold?)

Alert Overview

On May 8, 2025, analysts at the FPT SOC identified a cluster of outbound HTTP requests from several internal endpoints targeting the domain cdn.jsdelivrs[.]com. This domain, which resolves to IP 91.195.240.12, is registered under AS47846 (SEDO GmbH, Germany). Although seemingly benign at first glance, the domain closely resembles the legitimate jsDelivr CDN (cdn.jsdelivr.net) and triggered concern due to its typosquatted naming and non-standard usage patterns.

Initial review of DNS and proxy logs revealed browser-driven connections without any indication of direct user interaction with the domain itself, suggesting script-initiated activity embedded within a third-party webpage.

Investigation Details

1. Initial Analysis

Firstly, we found that all the affected hosts that accessed the same URL accessed the website sports[.]ltn[.]com[.]tw, a Taiwanese sports news website.

After checking the embedded JavaScript within the page invoked pop-up ads, one of which triggered the suspicious request to cdn.jsdelivrs[.]com.

App AnyRun Sandbox Result: https://app.any.run/tasks/90532748-66f1-4d09-a90e-d1ca27bd3f35

The flow appeared entirely browser-driven, with the endpoint process tree linking to chrome.exe, no CLI usage observed.

2. Domain Intelligence (WHOIS + Passive DNS)

AttributeValue
Queried Domaincdn.jsdelivrs[.]com
IP Address91.195.240.12
ASN / OrgAS47846 / SEDO GmbH
Passive DNS\>120 subdomain queries in past 180 days
WHOIS Created2017, tagged as monetized/parked domain
SSL CertificateSelf-signed, non-validating

The domain is a typosquatted clone with no legitimate CDN functions. Historical associations tie it to ad redirect platforms, SEO manipulation, and domain parking infrastructure.

3. Payload Characteristics

  • HTTP GET requests to cdn.jsdelivrs[.]com returned either empty responses or HTML with redirect scripts.

  • Redirection targets included sedoparking.com and similar monetization services.

  • No evidence of JavaScript obfuscation, encoded payloads, or exploit delivery.

  • DPI (where available) confirmed redirection to external parked domains, ~200 byte HTML responses, consistent with advertising scripts.

4. Correlation with jsDelivr Abuse Campaigns

While this specific domain was not directly tied to past phishing infrastructure, its pattern strongly resembles the attack flow documented in the 2023 Check Point report:

    • Malicious npm package (reactenz) was used to dynamically load an encoded HTML phishing page hosted on jsDelivr’s CDN.

      • jsDelivr retained the payload even after the source package (standforusz) was taken down from npm due to permanent caching policies.

      • The phishing flow was only visible at runtime due to obfuscation and dynamic DOM injection.

Assessment: In our case, no phishing payloads or encoded HTML were observed. However, the infrastructure setup (typosquatted domain, parked redirection) could be reused or repurposed with little effort by a motivated actor.

5. Traffic Flow Visualization

The diagram above outlines the interaction chain: a user visits a legitimate website, embedded JavaScript spawns an ad popup that connects to a typosquatted CDN-like domain, returning redirector content. A hypothetical branch illustrates the abuse pathway seen in phishing kits using similar delivery infrastructure.

Actions & Mitigations

Mitigation StepDetail
Network EnforcementMDE and firewall enforced a deny policy for all connections to the domain/IP
Threat Intel IntegrationDomain and IP added to the custom CrowdStrike IOC list
IOC HuntSIEM queried for the past 30 days across DNS/proxy logs for recurrence
Endpoint Process ReviewNo signs of persistence, scheduled tasks, or PowerShell/scripting activity
ATT&CK MappingConsidered T1566.002 (phishing via service), T1204.001 (user-triggered link)

📚 References