Training and CyberInsights Today

F5 ASM

17 lessons · Beginner to advanced

Policies, staging, learning, signatures, tuning, API security, logs and recovery.

F5 ASM training →

FortiWeb

17 lessons · Beginner to advanced

Server policies, profiles, ML, signatures, APIs, troubleshooting and HA.

FortiWeb training →

How to use the WAF training hub

CyberInsights Today

Daily cybersecurity analysis and practical bulletins

Understand recent vulnerabilities, threats, their impact and mitigation. When there is no worthwhile new issue, learn a security concept or defensive practice through clear explanations, illustrations and evidence.

Read CyberInsights Today →

OPTIMJACK and Optimizely: JavaScript Supply-Chain Incident Response

Download the OPTIMJACK response cheat sheet (PDF)

CYBERINSIGHTS TODAY · WEB SUPPLY CHAIN SECURITY

By Harpreet Singh · November 16, 2026

Summary. A website can serve harmful browser code even when its own release history looks clean. OPTIMJACK illustrates what happens when code delivered through a trusted experimentation platform changes outside the website owner's deployment process. Start by mapping which pages loaded the affected code and what visitors could do on those pages.

Incident updates · November 16, 2026

Report URI documented unauthorized JavaScript in sampled Optimizely Web snippets and reported cleanup across sampled projects later the same day. That observation does not establish a recovery time for every project. The reviewed evidence does not establish how publishing access was obtained.

1. Campaign scope and evidence

Optimizely documents that its experimentation snippet runs on pages where experiments are enabled. Report URI examined delivered snippets and reported cryptocurrency destination substitution, attempts to collect payment or other browser-readable information, and mobile exploit delivery. A count of contaminated snippets is a measure of sampled code. It cannot tell you how many visitors loaded it or how many lost data. The findings concern a campaign, and the sources reviewed do not establish a CVE or CVSS score. Report URI investigation · Optimizely snippet documentation

2. Where the browser trust boundary failed

Conceptual browser exposure path
  1. Experiment change outside the application release
  2. Trusted script delivery
  3. Code executes in the embedding page
  4. Page content or a transaction can be affected

HTTPS establishes whom the browser connects to. A script allowlist controls which sources it accepts. Both can succeed while the permitted source delivers changed code. A third-party script running in the embedding page can read or change content accessible there. A properly isolated cross-origin payment frame can protect its own fields, while the surrounding page can still be altered. If a visitor copied payment details, reconcile the actual destination against an independent transaction record.

3. Reconstruct the exposure window

  1. List project IDs, tag-manager rules, script URLs and page routes. Include regional paths, login, checkout and caches.
  2. Preserve copies of suspicious script bytes, hashes, timestamps, headers and publishing audit logs where feasible. Collect evidence without executing suspicious code.
  3. Compare delivered content against known reviewed versions. Look for unexpected destinations or changes in browser requests. Correlate with route and session evidence before assessing affected visitors.
  4. Separate code delivery, possible data access and confirmed impact. Record where evidence is missing and use transaction records to verify suspected payment redirection.

4. Contain and verify recovery

Defensive recovery workflow
  1. Preserve evidence and establish the window
  2. Disable affected delivery and secure publishing
  3. Restore a reviewed version and purge caches
  4. Verify browser requests and complete a synthetic transaction

Disable affected delivery on sensitive routes first and verify that visitors no longer receive it. Review privileged users and integrations, revoke suspicious sessions and rotate implicated credentials. Use Content Security Policy (CSP) to constrain executable scripts, outbound connections and frames. Trial the policy with reporting before enforcement. An allowed supplier can still deliver altered code. Subresource Integrity can verify exact bytes where the asset is immutable and the integration supports it. Frequently updated experimentation scripts need a reviewed promotion process before they can be pinned reliably.

Verify each critical route with a fresh session and, where relevant, an existing cached session. Record the delivered bundle identity, approved destinations and synthetic transaction result. Repeat after cache expiry and assign an owner who can stop delivery if unexpected changes return.

5. What could have been done better

Treat experiment publishing as production access. Require review for changes to sensitive routes, restrict publishing privileges and monitor what browsers actually receive. Rehearse disabling experimentation without interrupting critical customer flows. Keep payment confirmation independent of the page that displayed the destination.

Primary references

Comments

Popular posts from this blog

PHP2Shell Explained: Impact, Patches and Mitigation for CVE-2026-109618 and CVE-2026-109619

Start Here: F5 ASM Classroom — From Beginner to Advanced

Broken Object-Level Authorization (BOLA) in APIs: How to Find and Fix IDOR