OPTIMJACK Third-Party Script Supply Chain Impact
High severity TLP:CLEAR Research by Unit 42 Published 18 Nov 2026
Hypothesis
An adversary has exploited a trusted third-party vendor to distribute malicious JavaScript via their CDN, leading to mobile exploit delivery and credential skimming on internal corporate assets.
Brief
Why This Hunt Matters
The OPTIMJACK campaign, detailed by Unit 42 in "Borrowed Code, Real Consequences: What OPTIMJACK Teaches Us" (https://unit42.paloaltonetworks.com/optimjack-supply-chain-attack/), demonstrates the fragility of the third-party script ecosystem. Adversaries compromised the Optimizely CDN to inject malicious JavaScript into trusted pages. This script targeted users with mobile exploit kits, making the point of entry a legitimate, certificate-validated domain.
How the Hunt Flows
The hunt begins by scoping interactions with the Optimizely CDN within the hb_http_activity surface. This query establishes a candidate pool of hosts that loaded the potentially compromised script during the vulnerability window. By filtering for specific vendor domains, an analyst identifies the extent of exposure across the workstation fleet.
Next, the hunt pivots to triage evidence of the follow-on exploit stage. Two parallel queries search hb_http_activity and hb_dns_activity for keywords associated with the Coruna and DarkSword exploit kits. This phase looks for the actual delivery of the mobile payload or the resolution of attacker-controlled infrastructure.
In the final analysis phase, an analyst correlates the CDN interactions with the exploit kit markers. A host that visits the CDN and then immediately requests an exploit kit payload receives a malicious verdict. This correlation allows the hunt to find high-confidence compromise signals without alerting on every legitimate script load.
What the Hunt Cannot See
The hunt faces two primary visibility challenges. First, if the proxy gateway does not perform SSL/TLS decryption, full URL paths for exploit kit requests remain hidden. The hunt then relies on DNS telemetry, which may lack the granularity needed for certain exploit patterns. Second, the campaign targets mobile devices. If an employee uses an unmanaged iPad or iPhone that does not route traffic through corporate inspection, the exploitation stage remains invisible to internal telemetry.
Why This is a Hunt
This is a hunt because it correlates legitimate activity (CDN usage) with rare malicious follow-on activity across multiple surfaces. A single rule on Optimizely CDN traffic would be far too noisy to be actionable; an analyst must weigh the presence of exploit-kit markers against the CDN interaction to reach a verdict.
Steps
-
Query, scoping
Interactions with Optimizely CDN
Identify hosts that interacted with the Optimizely CDN during the vulnerability window to establish a candidate pool for script execution.
sql, reads hb_http_activitySELECT device_hostname, src_endpoint_ip, url_hostname, url_path, user_agent, MIN(time) AS first_access, MAX(time) AS last_accessFROM hb_http_activityWHERE instr(',' || '{{cdn_domains}}' || ',', ',' || LOWER(url_hostname) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days')GROUP BY device_hostname, src_endpoint_ip, url_hostname, url_path, user_agentWhat a hit looks like. Rows indicating which hosts visited the affected infrastructure. Silence means no recorded interaction with the CDN during the window.
-
Query, triage
Exploit Kit HTTP Traffic
Find HTTP requests containing keywords associated with the Coruna and DarkSword exploit kits.
sql, reads hb_http_activitySELECT device_hostname, src_endpoint_ip, url_full, user_agent, timeFROM hb_http_activityWHERE ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0) AND (instr(LOWER(url_full), 'coruna') > 0 OR instr(LOWER(url_full), 'darksword') > 0) AND time >= datetime('now', '-{{lookback_days}} days')What a hit looks like. Presence of exploit kit names in URL paths or hostnames. Silence indicates no clear HTTP markers for these kits were detected.
-
Query, triage
Exploit Kit DNS Resolutions
Search DNS history for infrastructure associated with the Coruna and DarkSword kits that may not have full HTTP URL logging.
sql, reads hb_dns_activitySELECT device_hostname, query_hostname, COUNT(*) AS count, MIN(time) AS first_seenFROM hb_dns_activityWHERE ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0) AND (instr(LOWER(query_hostname), 'coruna') > 0 OR instr(LOWER(query_hostname), 'darksword') > 0) AND time >= datetime('now', '-{{lookback_days}} days')GROUP BY device_hostname, query_hostnameWhat a hit looks like. DNS resolutions for domains including exploit kit keywords. Silence suggests no resolution of those kit-specific domains occurred.
-
Agent triage
Weigh CDN and Exploit Evidence
Assess whether hosts interacting with the Optimizely CDN subsequently triggered indicators for mobile exploit delivery.
-
Decision
Route on Verdict
Route the hunt to containment if malicious activity is confirmed, or to analyst review if the findings are ambiguous.
-
Response action
Isolate Impacted Host
Prevent potential lateral movement or further data exfiltration from a host showing exploit kit markers.
-
Analyst task
Detailed Analyst Investigation
Review the browser history, local storage, and logs on the affected host to confirm the extent of the JS injection impact.
-
Analyst task
Close Out Hunt
Record the findings and update the exclusion or detection logic for third-party script monitoring.
Coverage
2 of 4 stages covered
| Stage and ATT&CK technique | Covered | How, or why not |
|---|---|---|
| Malicious Script Distribution T1195.002 |
Yes | optimizely-cdn-interaction |
| Mobile Exploit Kit Delivery T1203 |
Yes | exploit-kit-http-traffic, exploit-kit-dns-resolutions |
| Vendor Platform Exploitation T1190 T1195 |
Out of scope | Exploitation of the vendor's internal platform is only visible in the vendor's own logs, not on the customer estate. |
| Client-Side Financial Collection T1539 T1557 |
Not visible | Wallet swapping and card skimming occur within the user's browser memory and are not typically captured in proxy or endpoint process logs unless exfiltration occurs to a known domain. |
Blind spots
- Needs SSL/TLS decryption on the proxy gateway. If the proxy only sees the destination IP/hostname due to encryption, specific exploit kit URI patterns will be missed, leading to false negatives. It would answer: Are the full URL paths for exploit kit requests visible?
- Needs Mobile Device Management (MDM) telemetry. The attack targets iOS and iPad devices. If these are not managed or do not route traffic through corporate inspection, the actual exploitation stage remains invisible. It would answer: Did the exploit kit successfully execute on personal mobile devices?
Parameters and data
Parameters
| Parameter | Type | Default | What it is |
|---|---|---|---|
cdn_domains | list[domain] | cdn.optimizely.com, optimizely.com, logx.optimizely.com | Legitimate vendor CDN domains that served the malicious script. |
lookback_days | number | 14 | Days of history to examine. |
scope_hosts | list[host] | none | Limit the hunt to specific hosts if needed; leave empty for fleet-wide. |
Data sources
| Source | Category | Telemetry |
|---|---|---|
| Endpoint telemetry (hb_ surfaces) | endpoint | endpoint |
| Web server / proxy logs | siem | network |
Source
An open hunt.md file. It runs anywhere that reads the format.
---
analysis: This is a hunt because it correlates legitimate activity (CDN usage) with
rare malicious follow-on activity across multiple surfaces. A single rule on Optimizely
CDN traffic would be far too noisy to be actionable; an analyst must weigh the presence
of exploit-kit markers against the CDN interaction to reach a verdict.
blind_spots:
- id: limited-proxy-visibility
question: Are the full URL paths for exploit kit requests visible?
requires: SSL/TLS decryption on the proxy gateway
risk: If the proxy only sees the destination IP/hostname due to encryption, specific
exploit kit URI patterns will be missed, leading to false negatives.
stage: mobile-exploit-kit-delivery
- id: unmanaged-mobile-exposure
question: Did the exploit kit successfully execute on personal mobile devices?
requires: Mobile Device Management (MDM) telemetry
risk: The attack targets iOS and iPad devices. If these are not managed or do not
route traffic through corporate inspection, the actual exploitation stage remains
invisible.
stage: mobile-exploit-kit-delivery
coverage:
- stage: malicious-script-distribution
status: covered
steps:
- optimizely-cdn-interaction
- stage: mobile-exploit-kit-delivery
status: covered
steps:
- exploit-kit-http-traffic
- exploit-kit-dns-resolutions
- reason: Exploitation of the vendor's internal platform is only visible in the vendor's
own logs, not on the customer estate.
stage: vendor-platform-exploitation
status: out_of_scope
- reason: Wallet swapping and card skimming occur within the user's browser memory
and are not typically captured in proxy or endpoint process logs unless exfiltration
occurs to a known domain.
stage: client-side-financial-collection
status: not_visible
guardrails:
claims: no_unsupported
evidence: citation_required
missing_data: not_benign
telemetry: untrusted
hunt:
applicability: campaign-specific
handoff: keep-as-periodic-hunt
justification: A negative result across the estate provides assurance that the OPTIMJACK
campaign did not lead to mobile exploit execution or financial data theft within
the enterprise, despite the use of trusted CDN infrastructure.
methodology: model-assisted
trigger: intel-report
hypothesis: An adversary has exploited a trusted third-party vendor to distribute
malicious JavaScript via their CDN, leading to mobile exploit delivery and credential
skimming on internal corporate assets.
labels:
- hunt
- attack.t1195.002
- attack.t1190
- attack.t1203
- collection
- execution
- initial access
name: OPTIMJACK Third-Party Script Supply Chain Impact
parameters:
cdn_domains:
default:
- cdn.optimizely.com
- optimizely.com
- logx.optimizely.com
description: Legitimate vendor CDN domains that served the malicious script.
from:
kind: article
observed: '2026-11-16'
ref: https://unit42.paloaltonetworks.com/optimjack-supply-chain-attack/
type: list[domain]
lookback_days:
default: '14'
description: Days of history to examine.
from:
kind: manual
observed: '2026-11-17'
ref: Standard lookback for recent supply chain compromise
type: number
scope_hosts:
default: []
description: Limit the hunt to specific hosts if needed; leave empty for fleet-wide.
type: list[host]
provenance:
authors:
- name: Huntbase hunt generation
org: huntbase.io
generated:
by: huntbase-hunt-generation
from: https://unit42.paloaltonetworks.com/optimjack-supply-chain-attack/
gates:
- dry-run
- lint
- critic
model: hb_google/gemini-3-flash-preview
rationale: Focus the hunt on administrative and developer workstations first, as these
roles often browse technical documentation or platforms that integrate Optimizely
tags. Widen the scope to all hosts with external browsing capability if the lead
query returns broad results.
references:
- name: "Unit 42 \u2014 OPTIMJACK Supply Chain Attack"
url: https://unit42.paloaltonetworks.com/optimjack-supply-chain-attack/
related:
- hunt: client-side-js-skimming-behavior
reason: Detecting the actual theft of card data or wallet swapping requires DOM-level
monitoring or specialized client-side telemetry not available in these surfaces.
relation: out-of-scope-alternative
scenario:
stages:
- name: Vendor Platform Exploitation
observables:
- Optimizely Web Experimentation platform vulnerability
- Project JavaScript field modification
- Nov 16, 2026
slug: vendor-platform-exploitation
tactic: initial-access
techniques:
- T1190
- T1195
- name: Malicious Script Distribution
observables:
- Optimizely CDN
- Legitimate vendor security certificate
- Malicious JavaScript snippets
- optimizely.com
slug: malicious-script-distribution
tactic: initial-access
techniques:
- T1195.002
- name: Client-Side Financial Collection
observables:
- Cryptocurrency wallet address swapping
- Payment card skimming
- Checkout form data collection
slug: client-side-financial-collection
tactic: collection
techniques:
- T1539
- T1557
- name: Mobile Exploit Kit Delivery
observables:
- Coruna exploit kit
- DarkSword exploit kit
- iOS devices
- iPad devices
slug: mobile-exploit-kit-delivery
tactic: execution
techniques:
- T1203
summary: The OPTIMJACK campaign involved the compromise of Optimizely's Web Experimentation
platform to inject malicious JavaScript into the websites of at least 178 organizations.
The tampered code, delivered via the vendor's legitimate CDN, was used to skim
payment card data, swap cryptocurrency wallet addresses, and deliver mobile exploit
kits to iOS users.
severity: high
targets:
analyst:
name: Tier-2 analyst
role: analyst
endpoint:
category: endpoint
name: Endpoint telemetry (hb_ surfaces)
telemetry:
- endpoint
hunter:
agent: true
name: Hunt agent
web:
category: siem
name: Web server / proxy logs
telemetry:
- network
tlp: clear
type: investigation
---
# OPTIMJACK Third-Party Script Supply Chain Impact
This hunt identifies hosts that interacted with the compromised Optimizely CDN infrastructure and subsequently engaged with known mobile exploit kit infrastructure (Coruna or DarkSword). By correlating trusted CDN traffic with rare exploit-delivery signatures across HTTP and DNS telemetry, the hunt uncovers potential data exposure even when delivered from a legitimate, certificate-validated source.
## optimizely-cdn-interaction
<!-- Interactions with Optimizely CDN -->
Identify hosts that interacted with the Optimizely CDN during the vulnerability window to establish a candidate pool for script execution.
```sqlite target=web role=scoping params=(cdn_domains=cdn_domains, lookback_days=lookback_days)
~~~yaml
expected: Rows indicating which hosts visited the affected infrastructure. Silence
means no recorded interaction with the CDN during the window.
reads:
- device_hostname
- src_endpoint_ip
- url_hostname
- url_path
- user_agent
- time
silence: not_evidence_of_absence
source: hb_http_activity
verified: dry-run
verified_at: '2026-11-18'
~~~
SELECT device_hostname, src_endpoint_ip, url_hostname, url_path, user_agent, MIN(time) AS first_access, MAX(time) AS last_access FROM hb_http_activity WHERE instr(',' || '{{cdn_domains}}' || ',', ',' || LOWER(url_hostname) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days') GROUP BY device_hostname, src_endpoint_ip, url_hostname, url_path, user_agent
```
## exploit-kit-activity-check
<!-- Check for Exploit Kit Activity -->
parallel:
- → exploit-kit-http-traffic
- → exploit-kit-dns-resolutions
join: → agent-triage
## exploit-kit-http-traffic
<!-- Exploit Kit HTTP Traffic -->
Find HTTP requests containing keywords associated with the Coruna and DarkSword exploit kits.
```sqlite target=web role=triage params=(scope_hosts=scope_hosts, lookback_days=lookback_days)
~~~yaml
expected: Presence of exploit kit names in URL paths or hostnames. Silence indicates
no clear HTTP markers for these kits were detected.
reads:
- device_hostname
- src_endpoint_ip
- url_full
- user_agent
- time
silence: not_evidence_of_absence
source: hb_http_activity
verified: dry-run
verified_at: '2026-11-18'
~~~
SELECT device_hostname, src_endpoint_ip, url_full, user_agent, time FROM hb_http_activity WHERE ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0) AND (instr(LOWER(url_full), 'coruna') > 0 OR instr(LOWER(url_full), 'darksword') > 0) AND time >= datetime('now', '-{{lookback_days}} days')
```
## exploit-kit-dns-resolutions
<!-- Exploit Kit DNS Resolutions -->
Search DNS history for infrastructure associated with the Coruna and DarkSword kits that may not have full HTTP URL logging.
```sqlite target=endpoint role=triage params=(scope_hosts=scope_hosts, lookback_days=lookback_days)
~~~yaml
expected: DNS resolutions for domains including exploit kit keywords. Silence suggests
no resolution of those kit-specific domains occurred.
reads:
- device_hostname
- query_hostname
- time
silence: not_evidence_of_absence
source: hb_dns_activity
verified: dry-run
verified_at: '2026-11-18'
~~~
SELECT device_hostname, query_hostname, COUNT(*) AS count, MIN(time) AS first_seen FROM hb_dns_activity WHERE ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0) AND (instr(LOWER(query_hostname), 'coruna') > 0 OR instr(LOWER(query_hostname), 'darksword') > 0) AND time >= datetime('now', '-{{lookback_days}} days') GROUP BY device_hostname, query_hostname
```
## agent-triage
<!-- Weigh CDN and Exploit Evidence -->
```agent target=hunter
cite: required
context:
- optimizely-cdn-interaction
- exploit-kit-http-traffic
- exploit-kit-dns-resolutions
max_iterations: 4
objective: Determine if any host visiting the Optimizely CDN also shows signs of exploit
kit delivery (Coruna/DarkSword) based on the HTTP and DNS telemetry.
success_criteria: A per-host verdict of malicious, suspicious, or benign based on
the sequence of events.
tools:
- endpoint
- web
```
## route-verdict
<!-- Route on Verdict -->
if~: "the agent-triage identifies at least one host with confirmed exploit kit traffic following a CDN interaction" (confidence: high, judge=hunter)
then: → isolate-impacted-host
indeterminate: → analyst-investigation
unavailable: → analyst-investigation (blind_spot: limited-proxy-visibility)
else: → close-out-hunt
## isolate-impacted-host
<!-- Isolate Impacted Host -->
```action target=endpoint
~~~yaml
approval: required
~~~
Isolate the host from the network immediately to prevent further exploitation and preserve the state for forensic imaging.
```
→ analyst-investigation
## analyst-investigation
<!-- Detailed Analyst Investigation -->
```manual target=analyst
Examine the host's browser logs for interaction with 'cdn.optimizely.com' and any subsequent redirects or requests to Coruna/DarkSword infrastructure. Check for signs of payment card data theft or mobile exploit downloads.
```
→ close-out-hunt
## close-out-hunt
<!-- Close Out Hunt -->
```manual target=analyst
Document the scope of the investigation, the hosts identified as interacting with the malicious CDN, and whether any compromise was verified. Review segmentation policies for third-party scripts.
```
→ end
Run this hunt against your own data.
In Huntbase, Scout runs every step across the tools you already connect, checks back through your history and returns a verdict with the evidence. The hunt.md file is open, so you can also take it anywhere that reads the format.
Drafted by machine (huntbase-hunt-generation, hb_google/gemini-3-flash-preview), checked by dry-run, lint, critic, then reviewed by a person.