# Optimizely Web Experimentation "OPTIMJACK" injection (16 Nov 2026): detection pack

**Pack version 1.0.0 (17 Nov 2026, sources checked 08:35 to 08:45 UTC).** An unauthorised party exploited a
vulnerability in Optimizely Web Experimentation and used it to add malicious JavaScript to some customer projects
(Optimizely's status notice, 17 Nov 2026 01:02 UTC). Every page that loaded an affected snippet ran a loader for a
payload called track.js: a crypto wallet address swapper, a card skimmer for first-party form fields and, on iPhones
and iPads, a loader for iOS exploit kits. No CVE is assigned and no score exists, so this pack does not call it
Critical: it is **Critical by impact (our framing)**, because it was exploited at scale through a trusted supply
chain. Optimizely says the vulnerability is fixed and the injected code removed; there is no customer patch.

> **Act now (OUR ADVICE).** Hunt DNS, proxy and endpoint data from 04:00 UTC on 16 Nov 2026 for the NEWS IOCs (KQL Q1
> to Q3, Sigma). Find where your sites and users load Optimizely (Q4 to Q4c). Save the snippets your sites serve and
> scan them with the checker. Bring iPhones and iPads to iOS 18.7.3 or later, or iOS 26.3 or later, and preferably the
> latest release (Q5, Q5b, checker `--ios`). If your own forms took card details on 16 Nov, treat those cards as
> possibly exposed.
>
> **Status of this pack.** The KQL in this pack has NOT been run against a live tenant. Neither has the converted
> Sigma in `sigma/converted/`. Test and tune them before you rely on them. The KQL parses cleanly against the table
> schemas we model, which is not the same as running it. Optimizely's status notice publishes no indicators, so this
> pack has no VENDOR IOC. Every indicator is a NEWS IOC labelled with its source (The Hacker News or
> BleepingComputer), which credit security researchers. Every other pattern is OUR PATTERN. Items not confirmed by a
> source are marked [UNVERIFIED].
>
> **KEV and due dates.** No CVE exists, so nothing here is in CISA's Known Exploited Vulnerabilities catalogue. Where
> KEV due dates exist, they bind US federal agencies only (BOD 26-04). Act now regardless.

## What happened (verified 17 Nov 2026)

| Item | What the sources say | Source |
|---|---|---|
| Vendor statement | "Optimizely identified a security incident in which an unauthorized party exploited a vulnerability in Experimentation and used it to add malicious JavaScript to some customer projects. We've fixed the vulnerability and removed the injected code from affected projects." Status "Monitoring", posted 17 Nov 2026 01:02 UTC; affects "Web Experimentation (Snippet Delivery)" | Optimizely status notice |
| Where the flaw was | Optimizely told BleepingComputer the vulnerability was in "the Experimentation account signup flow". The status notice itself says only "a vulnerability in Experimentation" | BleepingComputer (Optimizely statement) |
| Vendor actions | Status notice: secured affected user accounts and added protections against unauthorised code. Statement to BleepingComputer: blocked the vulnerable path, deployed a fix, removed the injected code, "ended active sessions, revoked tokens, and reset access for affected user accounts", blocked code delivery in Cloudflare, and "notified all affected customers" | Optimizely status notice; BleepingComputer |
| How | The attacker wrote a loader into the Project JavaScript of customer projects and left the rest of each snippet untouched; snippets were still served from cdn.optimizely.com with the vendor's own certificate. Sites serving the snippet from their own domain were affected too, because the content comes from the same project | The Hacker News |
| Loader versions | Three: most loaded 1688360[.]cc/track.js directly; one hid that address in base64 (decoded with atob()); one carried a full copy of the payload that runs if the domain is blocked | The Hacker News; BleepingComputer |
| Scale | Security researchers found the code in 188 of 1,611 snippets they checked, belonging to 178 organisations, and say the real number is likely higher. These are researcher figures reported by the press, not an Optimizely count; Optimizely has not published a number | The Hacker News; BleepingComputer |
| Payload | track.js swaps crypto wallet addresses (page text, forms, links, clipboard and outgoing requests). A card skimmer, added at 13:43 UTC, captures card number, expiry and CVV submitted through the site's own form fields and sends them to the attacker; it cannot read cards entered in a cross-origin payment iframe. Cookies (not HttpOnly ones) and a page screenshot are sent only after an address swap near wallet or payment wording | The Hacker News; BleepingComputer (card data sent to 1688360[.]cc/card.php) |
| iOS stage | On iPhones and iPads the code loads boot.js from two .icu domains, which picks an exploit chain by iOS version: Coruna (reported for iOS 13.0 to 17.2.1), DarkSword (reported for iOS 18.4 to 18.7) and, for iOS 26, a chain that only collects device details. The kit's own comments mark iOS 18.7.3 and later, and iOS 26.3 and later, as patched | The Hacker News |
| Device compromise | None confirmed. The stage was set up, but the press reports no iPhone or iPad actually taken over; researchers who retrieved it with an emulated browser describe "conditional delivery", "not a successful compromise of a physical iPhone or iPad" | The Hacker News |
| Crypto received | The attacker's wallets received about $7,040 by 19:25 UTC on 16 Nov (one estimate) or about $6,600 (another). This is not confirmed as theft from this campaign alone: two addresses were in use before the injections began, and "the chain does not show which transfers came from victims" | The Hacker News; BleepingComputer (wallet table) |
| Exploitation | Yes: the code was served to visitors on 16 Nov 2026. A cleaned snippet was reported re-infected with a hidden version, which indicates the attacker still had access at that point | The Hacker News |
| Injector history | The first injection loaded webshell[.]lol, described as the same actor's injector, which had also been planted through other third-party scripts since late the previous week and first seen on 17 Sep | The Hacker News |

### Timeline (UTC; London is on UTC today)

| Time | Event | Source |
|---|---|---|
| 16 Nov 04:21 | First injection seen, loading webshell[.]lol | The Hacker News |
| 16 Nov 13:43 | Card skimmer added to the payload | The Hacker News |
| 16 Nov 17:13 to 17:57 | Coruna iframe added to every visit from iOS 18.6 or earlier; those pages reload every 60 seconds to retry, a sign a visitor could notice | The Hacker News |
| 16 Nov 19:25 | Wallets had received about $7,040 (see the caveat above) | The Hacker News |
| 16 Nov about 20:40 | Mass clean-up began | The Hacker News |
| 16 Nov 21:52 | Only one of the 188 snippets still carried the code | The Hacker News; BleepingComputer |
| 17 Nov 01:02 | Optimizely status notice, "Monitoring" | Optimizely status notice |

### iOS fixed versions (checked against Apple)

The thresholds 18.7.3 and 26.3 come from the exploit kit's own comments as reported by the press. Apple's security
notes confirm both are real releases with fixes for exploited WebKit and dyld flaws. Apple does not name Coruna or
DarkSword, so the link between the kits and these CVEs is [UNVERIFIED].

| Release | Date | What Apple says | Source |
|---|---|---|---|
| iOS and iPadOS 18.7.3 | 12 Dec 2025 | WebKit CVE-2025-43529 and CVE-2025-14174: "Apple is aware of a report that this issue may have been exploited in an extremely sophisticated attack against specific targeted individuals on versions of iOS before iOS 26." | Apple security note 125885 |
| iOS and iPadOS 26.2 | 12 Dec 2025 | The same two WebKit fixes | Apple security note 125884 |
| iOS and iPadOS 26.3 | 11 Feb 2026 | dyld CVE-2026-20700, with the same exploitation wording; "CVE-2025-14174 and CVE-2025-43529 were also issued in response to this report." | Apple security note 126346 |
| Latest releases | 9 Nov 2026 | iOS 27.2.1, iOS 26.7.3 and iOS 18.7.12 | Apple security releases list |

On the 26.x line Apple's same-day sibling of 18.7.3 is 26.2, but the kit marks 26.3 as patched; the pack uses 26.3,
the stricter of the two. OUR ADVICE: move every device to the latest release its hardware supports.

### Indicators (NEWS IOC, defanged; the checker and KQL hold the live forms)

| Indicator | Role | Source |
|---|---|---|
| `1688360[.]cc` with `/track.js`, `/card.php`, `/hit.php` | Payload host; card data sent to /card.php | The Hacker News; BleepingComputer (domain, /track.js, /card.php) |
| `136.113.106[.]136` | The address 1688360[.]cc pointed to | The Hacker News |
| `webshell[.]lol` | Injector loaded by the first injection | The Hacker News |
| `5vobauua3ojpklx[.]icu`, `eil786npto9u5bq[.]icu` | iOS stage domains | The Hacker News |
| SHA-256 `f7b84b9a94f749f8ced3f0112d0a66894e8bbc3b4aa5592021d1b8976a9bad40` | Final version of track.js | The Hacker News |
| SHA-256 `617df364120adc47fb4f55dcc0c69a9b6c9e108a92c60bc2b64a266254ab02f3` | Earliest captured version of track.js | The Hacker News |
| Nine wallet representations (including a TON raw-format alias; ETH and BNB share one address) | Attacker wallets found in track.js | BleepingComputer (wallet table) |

The payload changed several times during the day, so the hashes are not exhaustive. The wallet addresses are in the
checker only (they appear in copies of track.js, not in network logs).

## Advice

**Optimizely's advice (credited to Optimizely):** the status notice gives no customer steps. Its statement to
BleepingComputer says affected customers have been notified and that updates will be published on
status.optimizely.com.

**What the press reports researchers advise (credited to security researchers, via The Hacker News and
BleepingComputer):** open the Project JavaScript in every project and the custom code in every experiment and look for
1688360, atob( or any script you did not write; review which users, API tokens and integrations can write to your
account and rotate their credentials; treat cards entered on your own forms on 16 Nov as possibly exposed; rotate
secrets kept in cookies without HttpOnly; deploy a strict Content Security Policy (allowing cdn.optimizely.com does
not help); ask Optimizely which project IDs changed and when; keep third-party tags off checkout pages and take cards
through the payment provider's cross-origin iframe; update iPhones and iPads to iOS 18.7.3 or later, or 26.3 or
later; visitors should check 16 Nov crypto payments and card statements. Researchers advise against broadly blocking
cdn.optimizely.com.

**OUR ADVICE (CybaVerse, not from Optimizely):**

Proactive:

1. Inventory every site and client that uses Optimizely Web Experimentation, including snippets served from your own
   domain (KQL Q4 to Q4c show only the cdn.optimizely.com form).
2. Save each snippet your sites serve and run the checker on the saved copies. Record what you scanned and when. A
   clean copy today does not prove the snippet was clean on 16 Nov.
3. Do not block cdn.optimizely.com broadly. Block the NEWS IOC domains and IP at DNS, proxy and firewall.
4. Content Security Policy: a strict `script-src` and `connect-src` that allows only the hosts each page needs, no
   `unsafe-eval`, and reporting turned on, so a new host like 1688360[.]cc is blocked and reported.
5. Payments: take card details only in the payment provider's cross-origin iframe or hosted page, and keep
   experimentation and other third-party tags off checkout pages.
6. Review Optimizely write access: users, API tokens, integrations and SSO; remove what is not needed, rotate the
   rest, and require MFA.
7. Mobile: bring managed iPhones and iPads to the latest release (at least 18.7.3 or 26.3) with Intune or your MDM
   (Q5, Q5b, checker `--ios`).

Reactive:

8. Hunt from 04:00 UTC on 16 Nov 2026 (Q1 to Q3, Q4b, Q6 and the Sigma rules). A hit establishes contact or a lookup, not necessarily execution or successful exploitation. Identify the site,
   user and device, and review whether malicious code ran and what data the page handled. Blocked traffic and
   security research can also generate hits.
9. Site owners whose snippet was affected: assume card numbers, expiry dates and CVVs entered in your own form fields
   on 16 Nov were exposed; rotate session tokens and any secrets held in cookies without HttpOnly; ask Optimizely
   which of your project IDs changed, when and how; keep your logs.
10. Notification (OUR ADVICE, not legal advice): consider telling affected customers, and take advice on whether this
    is a reportable incident for your acquirer or card brands under PCI DSS, and a personal data breach to report to
    the ICO within 72 hours under UK GDPR.
11. iPhones and iPads on 18.6 or earlier that visited an affected site on 16 Nov (a page reloading every 60 seconds is
    a tell): update them now and have them checked for signs of compromise. No device compromise has been confirmed.
12. Users who paid in crypto on an affected site on 16 Nov: compare the destination with the intended address. Users
    who paid by card: watch statements and consider replacing the card.

## Corrections to the teammate brief (checked against the sources)

* "Optimizely Web Experimentation flaw (in the signup flow, per press)": confirmed, with attribution. The status
  notice says only "a vulnerability in Experimentation"; "account signup flow" is from Optimizely's statement to
  BleepingComputer.
* 188 of 1,611 snippets and 178 organisations: confirmed as researcher figures reported by the press, not an
  Optimizely count. This pack names no affected organisation other than Optimizely, whose own main site and login
  page also carried the code (The Hacker News).
* Timeline 04:21 UTC to about 21:52 UTC, status "Monitoring" at 01:02 UTC on 17 Nov: confirmed. Mass clean-up began
  at about 20:40 UTC.
* "Skims card number, expiry and CVV": confirmed, but the skimmer was added at 13:43 UTC, not from the start.
* "Exfiltrates non-HttpOnly cookies and screenshots": confirmed, but only after an address swap near wallet or
  payment wording.
* Coruna (iOS 13 to 17.2.1) and DarkSword (iOS 18.4 to 18.7): confirmed as the press reports them (Coruna 13.0 to
  17.2.1). iOS 26 got a chain that only collects device details.
* "About $7k received": reported (about $7,040, or about $6,600 by another estimate), but not confirmed as theft from
  this campaign alone, because two addresses were already in use before the injections.
* "No confirmed device compromise": confirmed. The iOS stage was set up, but no iPhone or iPad is reported taken
  over.
* "60s page reload is a tell": confirmed, for the Coruna iframe added between 17:13 and 17:57 UTC for iOS 18.6 or
  earlier, not for every iOS visit.
* IOCs and both hashes: confirmed in The Hacker News indicator list. BleepingComputer confirms the domain, /track.js
  and /card.php, and adds wallet addresses (nine distinct representations, including a TON alias).
* iOS 18.7.3+ or 26.3+: confirmed as the kit's own "patched" marks (The Hacker News) and as real Apple releases
  (12 Dec 2025 and 11 Feb 2026). On the 26.x line Apple's sibling of 18.7.3 is 26.2; the pack keeps 26.3.
* "Search from 04:00 UTC 16 Nov": the advice reported in The Hacker News says back to 08:00 UTC; this pack uses
  04:00 UTC because the first injection was at 04:21 UTC.
* "Severity: Critical": no CVE or score exists; this pack says Critical by impact (our framing).
* "A checker that fetches a site's Optimizely snippet": our checkers never use the network. The checker scans copies
  you save yourself; one example save command is below, separate from the checker.

## Unverified items

* The link between Coruna or DarkSword and specific Apple CVEs (Apple names neither kit).
* That iOS 27 keeps the fixes (OUR ASSUMPTION; the checker and Q5 treat iOS 27 as fixed).
* Whether DarkSword's "18.7" upper bound means 18.7.0 only; the pack flags 18.7.1 and 18.7.2 as below fixed but not
  in the reported range.
* Table and field availability: DnsEvents, DeviceEvents DNS records, IntuneDevices field names, the iOS version
  format in DeviceInfo, and Defender network telemetry from iOS.
* What /hit.php does (listed by the source without a stated purpose).
* How long the vulnerability existed, how many visitors were exposed, and whether the attacker still has access.

## Hunts

| ID | Label | What it finds |
|---|---|---|
| Q1 | NEWS IOC (The Hacker News) | DNS lookups of the four indicator domains (DnsEvents, DeviceEvents) |
| Q1b | NEWS IOC (The Hacker News) | Endpoint connections to the indicator domains or IP (DeviceNetworkEvents) |
| Q2 | NEWS IOC (The Hacker News) | Proxy or firewall records for the domains, IP and the three payload paths (CommonSecurityLog) |
| Q2b | NEWS IOC (The Hacker News) | E-mail link clicks to the indicator domains (UrlClickEvents) |
| Q3 | NEWS IOC (The Hacker News) | Files with either published track.js SHA-256 (DeviceFileEvents, DeviceProcessEvents) |
| Q4 | OUR PATTERN | Devices loading cdn.optimizely.com since 04:00 UTC 16 Nov, flagged if they also contacted a NEWS IOC |
| Q4b | OUR PATTERN | A NEWS IOC connection within 10 minutes after the same device loaded cdn.optimizely.com |
| Q4c | OUR PATTERN | Proxy view of Q4 (CommonSecurityLog) |
| Q5 | OUR PATTERN [UNVERIFIED version format] | iOS and iPadOS devices below 18.7.3 or 26.3, with kit range (DeviceInfo) |
| Q5b | OUR PATTERN [UNVERIFIED table and field names] | The same on Intune inventory (IntuneDevices) |
| Q6 | OUR PATTERN [UNVERIFIED telemetry on iOS] | iOS or iPadOS devices that reached a NEWS IOC, with their version |

Sigma rules:

| Rule | Label |
|---|---|
| `dns_query_optimjack_news_ioc_domains.yml` | NEWS IOC (The Hacker News; BleepingComputer) |
| `net_connection_optimjack_news_ioc.yml` | NEWS IOC (The Hacker News) |
| `proxy_optimjack_news_ioc.yml` | NEWS IOC (The Hacker News; BleepingComputer) |
| `proxy_optimjack_icu_boot_js.yml` | OUR PATTERN (any .icu host serving /boot.js, generalised from the reported iOS stage) |

## Checker

`scripts/check-optimjack.sh` (bash) and `scripts/Check-Optimjack.ps1` (PowerShell 7) give identical output and exit
codes. They are strictly offline and read-only: no network access of any kind, no writes, no other programs beyond
grep, sha256sum, wc, find, sort, tr and cut (bash), and they never print file content, only file paths, labels, line
counts and the file's SHA-256. Paths are printed with non-printable bytes replaced by `?`.

**Saving a snippet (you run this yourself, separately; it is not part of the checker).** Copy the exact snippet URL
from the `<script>` tag in your page source (the path format is [UNVERIFIED] here, so do not guess it), then save it
from a machine allowed to fetch it, for example:

```
curl -sS -o snippet_site1.js "<SNIPPET_URL copied from your page source>"
```

Save every snippet you serve, including copies on your own domain, into one folder, and keep the response headers
if you can (one source advises recording Last-Modified).

**Running the checker:**

```
bash scripts/check-optimjack.sh --scan ./saved-snippets
bash scripts/check-optimjack.sh --scan snippet_site1.js --scan snippet_site2.js --extra-hash <64-hex SHA-256>
bash scripts/check-optimjack.sh --ios ios-inventory.csv
pwsh -File scripts/Check-Optimjack.ps1 -Scan ./saved-snippets -Ios ios-inventory.csv
pwsh -c "& ./scripts/Check-Optimjack.ps1 -Scan snippet_site1.js,snippet_site2.js"   # several paths
```

What `--scan` reports for each file (folders are searched recursively, symbolic links are not followed, files over
10 MiB are skipped):

* NEWS IOC (exit 1): any line containing `1688360`, `136.113.106[.]136`, `webshell[.]lol`, either .icu stage domain or a
  wallet address (case-insensitive), or a whole-file SHA-256 equal to either published track.js hash.
* OPERATOR (exit 1): a whole-file SHA-256 you pass with `--extra-hash` / `-ExtraHash`.
* OUR PATTERN (exit 1): the base64 form of `1688360[.]cc` at any of its three byte offsets (`MTY4ODM2MC5jY`,
  `E2ODgzNjAuY2`, `xNjg4MzYwLmNj`, case-sensitive), which catches the base64 loader variant.
* SEARCH TERM (exit 2): `atob(`, `.icu` or `track.js`, which the sources advise searching for but which also appear
  in legitimate code.
* OUR HEURISTIC (exit 2): H1 `atob()` of a quoted base64 literal of 24 or more characters; H2 `eval` or `Function`
  of `atob()`; H3 a script `src` set from `atob()`; H4 a file that creates a script element and also calls `atob()`.
  Review any match by hand against the code you wrote.

`--ios FILE` / `-Ios FILE` reads a CSV with the header exactly `device,os_version` (UTF-8 BOM and CRLF accepted;
blank and `#` lines skipped; plain CSV, no quoted fields). Versions are one to three numeric parts (missing parts
count as 0). Each row is `IOS BELOW FIXED` (17.x or older; 18.x below 18.7.3; 26.x below 26.3), `IOS AT OR ABOVE
FIXED` (18.7.3 or later on 18.x, 26.3 or later on 26.x, and 27.x as OUR ASSUMPTION [UNVERIFIED]) or `IOS UNKNOWN`
(19 to 25 and 28 or later, which are not released iOS versions, or a value that is not a version). Rows below fixed
also say whether the version is in the reported Coruna or DarkSword range and whether it is in the 18.6-or-earlier
group that got the 60-second reload iframe. A device name is printed only if it is 1 to 64 ASCII letters, digits,
dots, underscores or hyphens.

| Exit | Verdict | Meaning |
|---|---|---|
| 0 | CLEAN | No indicator, search term or heuristic in any file, and every iOS row at or above the fixed versions |
| 1 | IOC FOUND | A NEWS IOC string or hash, an OPERATOR hash, or the base64 form of the payload domain. Treat the snippet as compromised and follow the reactive steps |
| 2 | SUSPICIOUS | Only search terms or OUR HEURISTIC loader patterns. Review by hand |
| 3 | IOS BELOW FIXED | At least one iOS or iPadOS row below the fixed versions. Update those devices |
| 4 | UNKNOWN | A file skipped (link, unreadable, over 10 MiB), a folder with no files, or an iOS row that cannot be judged |
| 5 | USAGE OR READ ERROR | Bad option, missing path, bad `--extra-hash`, or an iOS CSV with a wrong header or no rows |

Precedence: 5 > 1 > 3 > 2 > 4 > 0.

## Contents

| Path | What it is |
|---|---|
| `kql/optimizely-optimjack-hunting.kql` | Q1 to Q6 (11 hunts) for Microsoft Sentinel and Microsoft Defender XDR |
| `sigma/*.yml` | 4 rules (3 NEWS IOC, 1 OUR PATTERN) |
| `sigma/converted/` | Kusto output for no pipeline, microsoft_xdr, microsoft_365_defender, sentinel_asim and azure_monitor |
| `scripts/check-optimjack.sh`, `scripts/Check-Optimjack.ps1` | Offline, read-only checkers with identical output and exit codes |
| `tests/` | Bash tests, Pester tests (including bash versus PowerShell parity), synthetic fixtures (example domains only, marked `FIXTURE_ONLY`), `run_all.sh`, `gates.sh` (static gates), `negative_gates.sh` (plants faults and checks the gates catch them), `kql_parse.ps1` (KQL parse check), `sigma_convert.sh` (regenerates `sigma/converted/`) |

### Sigma conversion (sigma-cli, pySigma Kusto backend)

`sigma check`: 0 errors, 0 condition errors, 0 issues. The network connection rule converts on every pipeline. The
DNS rule converts without a pipeline only (use KQL Q1). The two proxy rules convert on `sentinel_asim`
(`imWebSession`) and without a pipeline, and fail on the other three ("Unable to determine table name"); use KQL Q2.
Every failure is recorded in `sigma/converted/`.

## Testing

`tests/run_all.sh` runs shellcheck, PSScriptAnalyzer, the bash tests, the Pester tests (including bash versus
PowerShell parity and identical indicator constants), the static gates in `tests/gates.sh`, the KQL parse check, both
checkers inside a bwrap sandbox with no network and a read-only filesystem (with a sha256 check of the pack before and
after), `sigma check` plus a comparison of fresh conversions with `sigma/converted/` (`REGEN=1` rewrites them), and
then `tests/negative_gates.sh`, which plants one fault at a time in a private copy and fails unless the gates reject
every one. The gates check, among other things: no network, write or exec primitives in the checkers; fixtures marked
`FIXTURE_ONLY` with example domains and RFC 5737 addresses only and no live indicator; indicators identical in both
checkers, the KQL and Sigma, and defanged in this README; the hashes and the iOS thresholds as the sources state them;
labels on every hunt and rule; no em or en dashes; banned names checked by sha256; only the approved sources linked;
severity wording; the live tenant caveat; and every exit code documented. Indicator test files are built at run time
from split strings in a private temporary folder; they are inert text and are never fetched.


Last full run (17 Nov 2026): bash tests 114 passed, Pester 50 passed, static gates 21 passed, negative gates 48
caught and 0 missed, sandbox runs 40 passed (19 bash, 19 PowerShell, 2 sandbox checks), KQL 11 hunts with 0 parse
errors, Sigma 4 rules with 0 check issues and 10 documented conversion failures, shellcheck and PSScriptAnalyzer clean.

## ATT&CK mapping (MITRE)

T1195.002 Supply Chain Compromise: Compromise Software Supply Chain; T1189 Drive-by Compromise; T1185 Browser Session
Hijacking; T1539 Steal Web Session Cookie; T1565.002 Data Manipulation: Transmitted Data Manipulation (wallet address
swap); T1027 Obfuscated Files or Information (base64 loader); T1071.001 Application Layer Protocol: Web Protocols;
T1041 Exfiltration Over C2 Channel.

## Sources

Fetched 17 Nov 2026 from 08:35 UTC:

* Optimizely status, "Web Experimentation Security Incident" (posted 17 Nov 2026 01:02 UTC): https://status.optimizely.com/incidents/ftw0fh9wx43z
* BleepingComputer (16 Nov 2026), Optimizely's statement, the payload domain and paths, and the wallet table: https://www.bleepingcomputer.com/news/security/optimizely-supply-chain-attack-infects-major-brands-sites-with-malware/
* The Hacker News (17 Nov 2026), timeline, payload, iOS stage, advice and the indicator list: https://thehackernews.com/2026/11/hijacked-optimizely-scripts-served-card.html
* Apple security releases: https://support.apple.com/en-us/100100
* Apple, iOS and iPadOS 18.7.3: https://support.apple.com/en-us/125885
* Apple, iOS and iPadOS 26.2: https://support.apple.com/en-us/125884
* Apple, iOS and iPadOS 26.3: https://support.apple.com/en-us/126346
* MITRE ATT&CK: https://attack.mitre.org/techniques/T1195/002/ , https://attack.mitre.org/techniques/T1185/ , https://attack.mitre.org/techniques/T1539/ , https://attack.mitre.org/techniques/T1565/002/

In line with our policy, research is credited to security researchers only. We name no researcher or security firm,
and we link no exploit code.

## Changelog

* 1.0.0 (17 Nov 2026): first version. KQL Q1 to Q6, four Sigma rules, offline checkers with snippet scan and iOS
  inventory modes, tests and gates.
