Threat Actors often leverage domain-based infrastructure to host and facilitate malicious operations. When actors deploy these new domains, they often leave patterns that can be used to signature the infrastructure and link it to past known activity. 

Developing these signatures can be difficult, and there is little public documentation on how they can be performed. Today, we will look into a single domain indicator shared on X/Twitter and show you how to analyse it for patterns that lead to 36 additional domains. 

Our final analysis will review these domains and link them with high confidence to public reports on APT SideWinder. 


Before jumping in, here's a summary of what we'll cover in this blog.

  • Obtaining initial intelligence (Domain) From Twitter/X
  • Analysing the domain to find a pivot point
  • Pivoting on regular expressions, dates and domain registrars
  • Adjusting pivots to identify additional domains
  • Parsing data output with CyberChef and JPATH
  • Enriching output with WHOIS records
  • Establishing patterns in subdomains
  • Obtaining public intelligence reports to assist attribution
  • Using public reports to link activity to APT SideWinder

Initial Intelligence

Our initial investigation begins with a single domain indicator shared by DocGuard in a recent post on X. 

Note the domain's name of docs.mofa-services-server[.]top and consider that MOFA is an acronym for “Ministry of Foreign Affairs”. This will become important later. 

Our initial indicator is a domain, so we can begin with domain-based analysis, such as a passive DNS lookup.

The aim here is to obtain historical records of IP addresses to which the domain has resolved. We want to use the IP addresses to find other domains associated with the same IP infrastructure. 

Executing A Pivot On Our Initial Domain

Running a passive DNS lookup in SilentPush, reveals that the domain currently resolves to an IP address of 188.114.97[.]3 . This IP is hosted by CloudFlare on ASN 13335

CloudFlare is a commercial and "legitimate" reverse proxy for network infrastructure. In the context of Threat investigations, Cloudflare effectively acts as a way of anonymising infrastructure as the attackers “real” server is hidden behind the CloudFlare proxy.
Since CloudFlare proxies are often shared between thousands of unrelated customers. This complicates the analysis process and hence it is regularly utilised by malicious actors.

We can try to find related domains by performing a passive DNS lookup for 188.114.97[.]3, this will reveal any domains that have resolved to the same address. 

Below is the passive DNS lookup for 188.114.97[.]3, showing a large number of unrelated domains.

Our screenshot above reveals that 801666666 domains have resolved to the same address. 

As mentioned prior, this huge number of related domains is due to the usage of CloudFlare. We can narrow down the results by applying additional filters, but the number of results may still be in the 10’s of thousands. Hence, we attempted a similar pivot on the parent domain to establish any easier patterns. 

Since the parent domain is likely to be owned and controlled by the same actor, it can occasionally serve as a more accessible and more helpful pivot point.

Executing a Pivot On Our Parent Domain

Since pivoting on the initial docs subdomain had way too many results, we performed a similar lookup on the parent domain of mofa-services-server[.]top.

Parent domains aren’t always given the same protection as subdomains, and since they are typically controlled by the same actor, they serve as a far more helpful pivot point. 

The parent domain of mofa-services-server[.]top has only one known IP of 91.195.240[.]123, which is hosted on SEDO with ASN 47846, and was first seen on 2024-03-20.

A passive DNS lookup on this new IP 91.195.240[.]123 will allow us to determine any domains that have shared the same address. 

Performing this lookup identifies 770427 related domains. This is a huge number but significantly less than that of the original CloudFlare IP. 

Since this is still a vast number, we can leverage regular expressions to apply additional filtering to narrow down our results. Performed correctly, this can significantly reduce the number of related domains to a manageable number. 

(In this case, regular expressions will reduce the 770427 results down to only 6)

Building An Advanced Threat Intelligence Query

An advanced query allows us to apply specific filters that will significantly reduce the number of results. Before we can do this, we need to establish what exactly we will filter on.

Consider that we know the following information about mofa-services-server[.]top

  • It's hosted on 91.195.240[.]123
  • It uses a .top Top Level Domain
  • The domain name contains three words, separated by hyphens
  • The domain was first observed on 2024-03-20

An advanced query allows us to provide this information through date filters, network filters, and regular expressions.  The below parameters are how they can be applied in SilentPush.

  • 91.195.240[.]123 can be applied as a qanswer filter. 
  • TLD (.top) can be applied as \.top$ to the end of a domain_regex
  • Three words, separated by hyphens, can be applied as ^[a-z]{1,}\-[a-z]{1,}\-[a-z]{1,} at the beginning of the domain_regex
  • The first observed date can be applied as first_seen_after=2024-03-18 and first_seen_before=2024-03-22, this allows for +- 2 days of buffer on either side.

The complete regular expression used here is ^[a-z]{1,}\-[a-z]{1,}\-[a-z]{1,}\.top$ and if you are using SilentPush, the advanced query can be found in Advanced Query Builder -> PADNS Queries -> Live Unsanctioned Assets Lookup)

Below we can see what this looks like with the filters applied.

Applying these filters cuts the results down to only 7 domains. This is a great number and is significantly lower than the 770427 initially associated with the same IP 91.195.240[.]123. This means our filters were able to cut out 770420 results. 

The 7 resulting domains contain recurring “PK” (Pakistan) themes and common acronyms for Government agencies. 

Government themes were observed in our initial mofa-services-server[.]top indicator through mofa (Ministry of Foreign Affairs). 

Parsing JSON Data With CyberChef and JPATH

The results are returned in JSON format and contain a huge amount of information. We only need the resulting domains (for now), so we can use Python or CyberChef to extract the domain field. 

For the sake of simplicity, we leveraged CyberChef and a JPath expression to filter the JSON output to return the 7 resulting domains.

We achieved this with a JPath expression of response.records[*].query

(If you're unfamiliar with JPath, here is an excellent tutorial for understanding it)

The 7 resulting domains can be seen clearly below.

The 7 domains have a recurring theme of Pakistan and Government agencies. We can also observe a recurring theme of IT Support services through mentions of updates, server, download and services.  

(Later we'll see how these are TTPs of APT SideWinder)

  • PK - Shortening of Pakistan
  • NITB - National Information Technology Board
  • MOFA - Ministry of Foreign Affairs
  • Goverment - Misspelling of Government
  • Cabinet - Decision-making arm of the Government

These similarities indicate that the domains are related and that we’re onto something, especially given they share the same IP address and have close registration dates (as required by our filters)

So far, the domains share the same IP infrastructure, same naming schemes and similar registration dates. We can build on this and establish further commonalities, such as domain registrars, subdomains and associated files. 

Enriching Domains With WHOIS Records

One method we can use to establish further commonalities is to perform WHOIS lookups on the domains. A WHOIS lookup will provide information about who registered the domains and which domain registrar they were registered with.

If the same domain registrar and registration information can be seen across multiple domains, this can be an indication that the domains are related. 

Many services (such as WHOIS) can perform these lookups but are limited to individual searches. We will leverage SilentPush for our lookups, as it supports bulk searches and significantly speeds up our process. 

If you are using SilentPush, bulk lookups can be performed with Advanced Query Builder -> Enrichment Queries -> Domain Bulk

After exporting the resulting JSON and parsing it with CyberChef, we can see that 6/7 of the domains were registered with NameSilo on 2024-03-19 with exact registration times within minutes of each other. 

One of the resulting domains luxury-get-away[.]top features a different naming theme and registration time. For the purposes of this blog, we will ignore this domain for the remainder of this analysis. 

We now had 6 related domains, 5 of which were new and discovered through pivoting.

Consider that our analysis established these commonalities between the 6 domains. 

  1. Same theme of Government Entities and Pakistan
  2. Same naming pattern (3 words separated by hyphens)
  3. Same Top Level Domain of .top
  4. Same registration provider of NameSilo
  5. Same IP address 91.195.240[.]123 (and hence, the same ASN 47846)
  6. Same registration date 2024-03-19 and registration times between 04:09 and 04:11

Pivoting To Additional Domains

With 6 domains and 6 commonalities between them, we had a strong base to begin performing additional pivots.  

We expanded our search by making the following adjustments to our query. 

  • Relaxing the IP requirement to allow for any IP hosted on SEDO/47846,
  • Requiring NITB, PK or Pakistan keywords in our regex
  • Relaxing the date requirements to any time after 2024-03-01
  • Adding a registrar requirement of NameSilo
  • Relaxing the TLD requirements  (Any TLD, not just .top)

Applied as an advanced search in SilentPush, this translated to

  • Pakistan keywords, domain_regex=^([^\.]{1,}\-|)(nitb|pk|paki(stan|))(\-[^\.]{1,}|)\.[a-z]{1,}$
  • Registered with NameSilo registrar=NameSilo
  • Registered after March 2024 whois_date_after=2024-03-01
  • Hosted on SEDO/47846, asnum=47846
  • Hosted on any Top Level Domain \.[a-z{1,}$ (located at the end of our domain_regex)
In SilentPush, this is an advanced domain search which can be found under Advanced Query Builder -> Domain Queries -> Search.

Also, we'd like to extend a special thanks to the SilentPush research team who helped with this section.

After running the search and filtering the JSON output, we obtained five results.

Two of these results were not present in our initial search. This meant that we now had a total of 8 domains related to our initial indicator.

The two new domains were largely thanks to our relaxation of the Top-Level-Domain requirements.

The new domains correspond to

  • Govt-pk[.]com
  • Moma-gov-pk[.]org

We now had three additional keywords that we could use to expand our search again. These keywords correspond to gov, govt and moma (Ministry of Maritime Affairs)

We then executed another advanced domain search, identical to before but now with a new regular expression.

  • Government Themes in domain name, domain_regex=^([^\.]{1,}\-|)(gov(t|erment|)|moma)(-[^\.]{1,}|)\.[a-z]{1,}$
  • Registered with NameSilo registrar=NameSilo
  • Registered after March 2024 whois_date_after=2024-03-01
  • Hosted on SEDO/47846, asnum=47846
  • Hosted on any Top Level Domain \.[a-z{1,}$

Updating the previous query to this regex returned 11 results. These results continued the theme of government entities, with references to 

  • NCSC (Possibly National Centre for Cyber Security)
  • MOHRE (Ministry of Human Rights)
  • IRS (Internal Revenue Service)
  • PPY (possibly Pakistan Patriotic Youth)

The results demonstrated a new recurring naming theme of govpk and gov-pk, so we again modified the regular expression to target these domain themes.

  • Gov/Gov-pk keywords in the domain name, domain_regex=^([^\.]{1,}\-|)(gov\-pk|govpk)(-[^\.]{1,}|)\.[a-z]{1,}$
  • Registered with NameSilo registrar=NameSilo
  • Registered after March 2024 whois_date_after=2024-03-01
  • Hosted on SEDO/47846, asnum=47846
  • Hosted on any Top Level Domain \.[a-z{1,}$ (end of domain_regex)

Executing this new query returned 3 results, two of them are new and feature government entities listed below.  

  • Paknavy (Pakistan Navy)
  • DGPS (Directorate General Ports and Shipping)

So far, all of our searches have required the domain to be associated with an ASN number of SEDO/47846.

One method we used to identify additional domains was to relax the ASN requirement whilst maintaining the other parameters.

We repeated this for all of our searches, with one in particular bringing in new results. 

  • Any domain name containing gov ,gov-pk or govpk and hyphens domain_regex =^([^\.]{1,}\-|)(gov\-pk|govpk)(-[^\.]{1,}|)\.[a-z]{1,}$
  • Hosted on any Top Level Domain (The final \.[a-z]{1,}$ in our regex)
  • Hosted on any Autonomous System (removal of the asnum parameter)
  • Registered with NameSilo
  • Registered after 2024-03-01

Executing the search with new parameters provided 9 domains, multiple of which were new and continued the theme of impersonating Pakistan Government entities.

  • EP  - Express Mail Track & Trace System
  • NADRA - National Database and Registration Authority
  • PTA - Pakistan Telecommunication Authority
We can continue to use the same concept of relaxing and adjusting parameters to identify additional domains. However, to keep this post from getting too long (there are infinite possible pivots), we’ve decided to leave the pivoting section here and continue with our currently identified domains. 

Results from our additional pivots will be included in the final IOC section of this post

Establishing Patterns in Subdomains

Recall that the initial domain shared by DocGuard had the primary malicious activity under the docs subdomain of docs.mofa-services-server[.]top

We wanted to see if our new domains had any such subdomains which could establish a further pattern linking the activity to the initial domain.

Recall the docs.mofa-services-server[.]top domain shared by DocGuard. The docs subdomain was first seen on 2024-05-01 , approximately 6 weeks after the parent domain was first registered.

If you are using SilentPush, this search can be found under Advanced Query Builder -> PADNS Queries -> Subdomain Records

We ran an identical search for our documents-server-pk[.]top domain, which revealed a similar pattern where a pmo subdomain was created approximately 6 weeks after the parent domain first appeared.

One theory is that the Threat Actor is “sitting” on parent domains and then performing malicious activity via subdomains at a later date. This may be to avoid domain-based filtering that blocks or alerts on recently registered infrastructure (<30 days old) . 

Repeating the subdomain searches returned a total of 15 subdomains featuring Government themes and new government entities of

  • MOD (Ministry of Defense)
  • ECP (Election Commission of Pakistan)
  • CTD (Counter Terrorism Department)
  • LGCD (Local Government and Community Development)
  • PUBAD (Ministry of Public Administration, Home Affairs and Provincial Councils). 

We can see these themes in the screenshot below.

Of additional interest here is that we see domains targeting Sri Lanka (lk) and Nepal, and that the majority of subdomains exist under gov-pk[.]com, which is an impersonation of the legitimate domain gov[.]pk

We can also observe that[.]com is an impersonation of the legitimate Sri Lankan domain[.]lk

Most of the identified parent domains did not have an associated subdomain. We believe this is likely due to the “waiting” that the actor is using after the parent domain is first created. 

At the time of this writing, it had only been 6 days since the first malicious subdomains were observed. Hence, we believe that the remaining subdomains had not yet been created. 

Linking Domains to APT SideWinder

At the time of this writing, we could not find any publicly available reports on our 37 newly identified domains. 

However, we found two extremely interesting reports by BlackBerry and Group-IB that detail 2023 activity of the Indian Advanced Persistent Threat (APT) known as SideWinder. This Threat Actor is known for targeting Pakistan, Nepal and Sri Lanka. (All in line with the activity we observed so far)

Both reports provide the following details and TTP’s regarding the SideWinder group. 

  • Primary targeting of South Asian countries bordering India
  • Heavy usage of domains impersonating Government Entities
  • Heavy targeting of Military and Government Entities
  • Heavy usage of Initial Access via Weaponized Documents with Government Themes

Domain Similarity, Government Entities and Primary Targeting of South Asian Countries 

The BlackBerry report contains a list of known SideWinder domains targeting South Asian countries.

The following domains were extracted from the BlackBerry report and show remarkable similarities to those identified during our analysis. Note the heavy usage of...

  • Hyphens in domain names
  • Recurring themes of Government entities
  • Recurring themes of Pakistan and Sri Lanka
  • Heavy usage of Subdomains

Although we have grouped this under one heading, this screenshot represents 4 unique commonalities between the domains we identified and known activity from APT SideWinder. 

The second report from Group-IB shows similar domains with remarkable similarities to those identified in our analysis. 

Public Reports of Initial Access Via Weaponized Documents

Both the Group-IB and BlackBerry reports detail SideWinder activity where initial access is achieved via weaponised documents with Government Entity themes. 

Additionally, both reports detail a malicious document titled GUIDELINES FOR BEACON JOURNAL – 2023 PAKISTAN NAVY WAR COLLEGE (PNWC).doc

The reports detail that this document leveraged a remote template injection vulnerability CVE-2017-0199 to download a remote file named file.rtf that contained obfuscated Javascript code. 

A visual overview of the document (Taken from BlackBerry and Group-IB) can be seen below. 

The BlackBerry article details another SideWinder document featuring Pakistan Government themes and an overall well-made and professional-looking email. 

Weaponized Documents In Our Newly Identified Domains

Reviewing one of our discovered domains paknavy-govpk[.]info on VirusTotal, we can see an associated .docx file named MoITT_federalemp[.]docx from 2024-04-29

This file, which is linked to our new domains, features the same CVE-2017-0199 exploit as detailed by BlackBerry and Group-IB

By searching the file hash 7dca552bc38f54716c80eb2c4f1f35cf6e5b12a78a5cec8bf335453c1b433cfd on Hybrid-Analysis, we noticed that the document contained Pakistan government themes and an overall similar structure to the publicly reported documents.

Below is the document associated with our domain paknavy-govpk[.]info

The usage of Pakistan government themes, weaponized documents and the same exploit CVE-2017-0199 is a strong indication that this activity is linked to the public SideWinder reports. 

Modification of Weaponized Documents

We could not find any additional presence of weaponised documents leveraging CVE-2017-0199 in our remaining domains. Although, we did observe a change to PDFs linking to password-protected .zip files. 

This represented a change in specific tactics, but continuing the overall tactic of weaponized documents.

The initial domain docs.mofa-services-server[.]top is related to the following document (shared by DocGuard in their initial post), which features a password-protected .zip with a password of pmo@letter

One of our identified subdomains pmo.documents-server-pk[.]top is related to a similar PDF file linking to a .zip.

By taking the hash from VirusTotal and searching it on Hybrid-Analysis, we see a similar theme of Government entity-themed phishing with password-protected .zip files.

The end of the document featured a prompt to download a password-protected file.

The presence of password-protected .zip files (likely containing malware) instead of CVE-2017-0199 represents both a strong link (via weaponized docs) and a slight change in SideWinder activity and techniques.

The overall tactic of weaponized documents is continued, but the specific tactic of CVE-2017-0199 has changed to a password-protected zip file.

SideWinder Usage of NameSilo

A subset of the older SideWinder domains shared by BlackBerry and Group-IB feature NameSilo as the domain registrar.

Many shared domains did not feature NameSilo, but this shows that SideWinder is familiar with NameSilo and uses it for a subset of their domain infrastructure.

Since all of the domains we featured today utilised NameSilo, this indicates a weaker but still useful connection between the new domains and those already publicly attributed to SideWinder.


We have now analysed a single domain indicator with threat intelligence tooling and identified 37 new domains with strong relations to known SideWinder activity. We analysed historical records around IP addresses, domain registrars, registration dates, associated files, and subdomains.

The tool used in this analysis was SilentPush, If you'd like to follow along, consider signing up for the Community Edition.

Domain Indicators