At The Edge Clear: August 10 - 17, 2026

Table of Contents
Loading nav...

‍

At The Edge is GreyNoise's weekly intelligence brief produced exclusively for customers incorporating complete IOCs, infrastructure attribution, detection guidance, and role-based recommendations. At The Edge Clear is a preview highlighting a couple insights and is available to the public.

‍

The Exposure Inventory Is the Weak Link

Analysis Period: August 10 – August 17, 2026

Every finding this period lands on an asset organizations expose to the internet deliberately and do not fully inventory: a gateway appliance owned by another team, a legacy application server, a relocated service, and a forward proxy with no identified owner. Adversaries attempted to exploit CISA KEV-listed flaws across five remote access and gateway vendors. Ten rented hosts each swept a disjoint contiguous band, together tiling the low and high ends of the TCP port range, and ninety-two more ran a matched toolkit that tests whether a discovered proxy forwards traffic.

‍

By The Numbers:

  • 5 vendors β€” Remote access and gateway vendors adversaries tried to exploit.
  • 17 β€” ColdFusion tags spanning 2009 to 2026.
  • ~5 days β€” For 10 hosts to sweep their disjoint port bands.
  • 82% β€” Of all Open Proxy Scanner and SOCKS5 Proxy Scanner probes, from 92 hosts.

‍

Preview Findings:

1. Five vendors, all nine flaws already KEV-listed

Adversaries attempted to exploit KEV-listed flaws across Ivanti, Palo Alto Networks, Citrix, F5 and Fortinet, several listed months ago. Ordering this work by observed volume runs backwards: adversaries made double-digit attempts against the maximum severity flaw while scanning the same product classes hundreds of thousands of times. Rank by KEV listing and reachability instead.

2. Seventeen ColdFusion tags, one likely module list

Adversaries attempted to exploit Adobe ColdFusion across seventeen distinct GreyNoise tags in one window, spanning 2009 to 2026 and including a maximum severity path traversal added to the KEV catalog in July. No single tag dominates the distribution, which is consistent with one module list run against every instance a source finds.

3. Ten hosts split the low and high ends of the port range

Ten rented hosts each took a disjoint contiguous band of the TCP port range, together covering its low and high ends. GreyNoise observed none of them before 12 August, and all ten presented a single TCP client fingerprint. They enumerated a service relocated to a port inside those bands on the same schedule as one sitting on its default.

4. Ninety-two hosts test whether a proxy really forwards

Two hosting networks ran a matched toolkit against unauthenticated web proxies and SOCKS5 relays, producing 82% of the Open Proxy Scanner and SOCKS5 Proxy Scanner probes GreyNoise observed this period. The two scanners ran within 1% of each other on both networks, on separate connections rather than one probe tripping two tags. Audit each proxy you operate from an external address and confirm the service refuses the relay.

‍

‍

Want the full brief?

GreyNoise customers get detailed briefs with complete IOCs, infrastructure attribution, detection guidance, and role-based recommendations every week.

‍Request a demo to learn more about GreyNoise's data and intelligence.

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

Read the transcript

‍

At The Edge is GreyNoise's weekly intelligence brief produced exclusively for customers incorporating complete IOCs, infrastructure attribution, detection guidance, and role-based recommendations. At The Edge Clear is a preview highlighting a couple insights and is available to the public.

‍

The Exposure Inventory Is the Weak Link

Analysis Period: August 10 – August 17, 2026

Every finding this period lands on an asset organizations expose to the internet deliberately and do not fully inventory: a gateway appliance owned by another team, a legacy application server, a relocated service, and a forward proxy with no identified owner. Adversaries attempted to exploit CISA KEV-listed flaws across five remote access and gateway vendors. Ten rented hosts each swept a disjoint contiguous band, together tiling the low and high ends of the TCP port range, and ninety-two more ran a matched toolkit that tests whether a discovered proxy forwards traffic.

‍

By The Numbers:

  • 5 vendors β€” Remote access and gateway vendors adversaries tried to exploit.
  • 17 β€” ColdFusion tags spanning 2009 to 2026.
  • ~5 days β€” For 10 hosts to sweep their disjoint port bands.
  • 82% β€” Of all Open Proxy Scanner and SOCKS5 Proxy Scanner probes, from 92 hosts.

‍

Preview Findings:

1. Five vendors, all nine flaws already KEV-listed

Adversaries attempted to exploit KEV-listed flaws across Ivanti, Palo Alto Networks, Citrix, F5 and Fortinet, several listed months ago. Ordering this work by observed volume runs backwards: adversaries made double-digit attempts against the maximum severity flaw while scanning the same product classes hundreds of thousands of times. Rank by KEV listing and reachability instead.

2. Seventeen ColdFusion tags, one likely module list

Adversaries attempted to exploit Adobe ColdFusion across seventeen distinct GreyNoise tags in one window, spanning 2009 to 2026 and including a maximum severity path traversal added to the KEV catalog in July. No single tag dominates the distribution, which is consistent with one module list run against every instance a source finds.

3. Ten hosts split the low and high ends of the port range

Ten rented hosts each took a disjoint contiguous band of the TCP port range, together covering its low and high ends. GreyNoise observed none of them before 12 August, and all ten presented a single TCP client fingerprint. They enumerated a service relocated to a port inside those bands on the same schedule as one sitting on its default.

4. Ninety-two hosts test whether a proxy really forwards

Two hosting networks ran a matched toolkit against unauthenticated web proxies and SOCKS5 relays, producing 82% of the Open Proxy Scanner and SOCKS5 Proxy Scanner probes GreyNoise observed this period. The two scanners ran within 1% of each other on both networks, on separate connections rather than one probe tripping two tags. Audit each proxy you operate from an external address and confirm the service refuses the relay.

‍

‍

Want the full brief?

GreyNoise customers get detailed briefs with complete IOCs, infrastructure attribution, detection guidance, and role-based recommendations every week.

‍Request a demo to learn more about GreyNoise's data and intelligence.

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍