Microsoft Uncovers Massive npm Supply Chain Attack Targeting CI/CD Pipelines Through Compromised @antv Packages

Listen to this Post

Featured Image

Introduction

Software supply chain attacks have become one of the most dangerous threats facing modern development environments. Attackers are no longer focusing only on breaking into networks directly. Instead, they are targeting trusted software dependencies that developers install every day without suspicion.

Microsoft has now revealed a major supply chain compromise involving the widely used npm ecosystem. A threat actor successfully breached the maintainer account behind @antv, a popular JavaScript visualization package organization, and weaponized trusted packages to distribute malware across downstream projects.

The attack did not stop at a single package. Through dependency chains, malicious code spread into development pipelines, cloud infrastructure, and continuous integration environments, creating a potentially enormous blast radius that affected thousands of software projects worldwide.

Attack Began With a Trusted Package Compromise

According to

After gaining access, malicious package versions were published directly into npm repositories. Because software developers commonly rely on automated dependency updates, the malicious code rapidly propagated downstream.

One particularly significant downstream victim was echarts-for-react, a package receiving more than one million weekly downloads. The compromise expanded beyond individual developers and entered CI/CD environments, cloud workloads, and automated build systems.

The malicious component itself was a heavily obfuscated JavaScript payload roughly 499 KB in size. Its design focused specifically on stealing credentials and compromising GitHub Actions environments.

The attack chain followed a structured pattern:

Maintainer account takeover

Publication of malicious package versions

Dependency-chain amplification into downstream software

Automatic execution through npm installation hooks

Multi-stage payload deployment using Node.js, shell execution, Bun runtime installation, and secondary malware components

The malware silently activated during package installation using npm lifecycle scripts, making detection particularly difficult.

Highly Sophisticated Obfuscation Techniques

Researchers discovered multiple layers of concealment designed to evade both automated security tools and manual analysis.

The first layer relied on over 1,700 Base64-encoded strings hidden inside rotated arrays. Critical operational values were concealed using runtime decoding mechanisms.

A second protection layer encrypted important values such as command-and-control infrastructure and environment variables using custom PBKDF2 combined with SHA-256 decryption logic.

The malware also contained environmental awareness features.

If execution occurred outside GitHub Actions or outside Linux-based systems, the payload exited immediately. This reduced visibility and limited exposure during security analysis.

Additional stealth measures prevented interaction with protected branches including:

main

master

dependabot

renovate

gh-pages

These safeguards reduced the likelihood of detection inside source repositories.

Credential Theft Across Multiple Platforms

The malware targeted an unusually broad collection of services and cloud environments.

GitHub credentials represented a primary objective. Attackers harvested GitHub tokens, searched for personal access credentials, validated account permissions, and attempted repository and organizational secret extraction.

Amazon Web Services environments also faced heavy targeting. The malware accessed instance metadata services, Elastic Container Service metadata endpoints, AWS configuration files, environment variables, and Secrets Manager resources across cloud regions.

HashiCorp Vault systems were another focus. The malware scanned more than a dozen locations for stored Vault tokens while attempting connections to local Vault services.

npm authentication tokens faced direct theft attempts as well. Attackers validated package publishing permissions and attempted token exchanges for elevated publishing capabilities.

Kubernetes secrets were harvested through service account token extraction and namespace enumeration.

Even password management infrastructure was targeted. Researchers observed attempts to interact with 1Password command-line tooling and efforts aimed at credential extraction.

The breadth of credential theft capabilities demonstrated careful planning and extensive knowledge of enterprise development environments.

GitHub Actions Runner Memory Scraping Raises Concern

One of the most alarming capabilities involved GitHub Actions runner memory extraction.

Rather than relying solely on environment variables or stored secrets, the malware scanned Linux process memory directly.

Attackers identified GitHub Actions worker processes and extracted runtime secrets directly from memory.

This approach bypassed standard secret masking protections.

Even credentials protected during logging or pipeline execution could become exposed if resident inside process memory.

CI/CD environments increasingly serve as gateways to production infrastructure. Compromising them often creates access to repositories, cloud systems, deployment credentials, and organizational secrets simultaneously.

That makes attacks like this exceptionally dangerous.

Privilege Escalation and Persistence

The malware did not stop at credential theft.

Researchers observed attempts to establish elevated system privileges by modifying sudo configurations.

DNS manipulation capabilities were also present through /etc/hosts changes, potentially allowing traffic redirection or infrastructure impersonation.

Persistence mechanisms extended further through:

Deployment of token monitoring processes

Bun runtime installation for second-stage payload execution

Repository enumeration for lateral movement

Expansion into additional developer assets

The malware also attempted something particularly troubling.

It forged Supply Chain Levels for Software Artifacts (SLSA) provenance attestations using Sigstore infrastructure.

Supply chain verification systems exist to increase trust.

Forging provenance undermines confidence in security validation mechanisms themselves.

Multi-Layer Data Exfiltration Infrastructure

Researchers identified several exfiltration methods.

Primary communications occurred through encrypted HTTPS command-and-control channels.

If direct infrastructure failed, attackers switched to Git Data APIs, creating repository blobs, trees, and commits to hide stolen information inside source control systems.

A third fallback mechanism created public repositories under victim accounts.

Researchers observed over 2,200 suspicious repositories connected to campaign activity.

These repositories carried unusual reversed descriptions that helped investigators identify attack artifacts.

Multiple exfiltration paths ensured operational resilience.

Even if defenders blocked one channel, attackers retained alternatives.

GitHub Responded Quickly

GitHub moved rapidly after learning about the campaign.

Actions included:

Removal of 640 malicious packages

Invalidation of 61,274 npm access tokens with elevated permissions

Dependabot alerts for affected users

npm audit notifications

Security advisory publication

Ongoing ecosystem monitoring

Authors behind affected repositories have since confirmed that remediation efforts resolved the immediate package compromise.

However, organizations exposed during the attack window remain at risk.

Mitigation Guidance From Microsoft

Microsoft recommends organizations immediately review dependency trees for affected packages.

Security teams should identify systems that installed compromised versions and rotate potentially exposed credentials.

Recommended actions include:

Pin trusted dependency versions

Disable npm lifecycle scripts using npm install –ignore-scripts

Rotate CI/CD secrets

Replace npm authentication tokens

Audit GitHub repositories for unexpected changes

Review package lockfiles and build artifacts

Inspect outbound network activity

Investigate suspicious package execution behavior

Enable endpoint protection tooling

Review cloud environments for compromised credentials

Organizations using developer environments connected to cloud infrastructure should treat exposure seriously.

Credential theft inside CI/CD systems frequently leads to broader infrastructure compromise.

What Undercode Say:

This incident highlights a major shift in attacker priorities. Traditional endpoint attacks remain important, but software supply chain compromise increasingly delivers higher returns for threat actors.

Modern development ecosystems rely heavily on trust.

Developers install packages assuming maintainers remain secure.

Automation tools continuously update dependencies.

CI/CD systems store secrets required for deployment.

Cloud infrastructure increasingly connects directly into build environments.

Attackers understand these realities.

Compromising one trusted maintainer account can create downstream access into thousands of organizations simultaneously.

What makes this campaign particularly concerning is operational maturity.

This was not opportunistic malware.

The attackers implemented layered obfuscation.

They built environment detection safeguards.

They deployed credential harvesting across multiple platforms.

They implemented fallback exfiltration channels.

They targeted CI/CD systems specifically.

They even attempted provenance forgery.

That reflects planning associated with advanced supply chain operations.

Another major takeaway involves developer convenience versus security.

Automated dependency upgrades accelerate productivity.

Lifecycle installation scripts simplify package deployment.

Build systems centralize credentials to streamline pipelines.

Every convenience feature expands attack surface.

Organizations may need stronger controls around dependency validation and package trust.

Software Bills of Materials (SBOMs), package pinning, isolated build environments, and stricter CI/CD credential segmentation become increasingly important.

Memory scraping techniques observed here also deserve attention.

Security teams often focus on protecting secrets at rest.

This attack targeted secrets in use.

Runtime credential protection remains an evolving challenge.

Cloud-native organizations should pay particular attention.

Compromised CI/CD environments frequently create pathways into production systems.

Attackers increasingly recognize development pipelines as high-value infrastructure.

Supply chain security frameworks like SLSA remain valuable, but this campaign demonstrates trust systems themselves may become targets.

Security models cannot rely solely on verification mechanisms.

Defense-in-depth remains critical.

Organizations should also revisit dependency hygiene.

Many companies do not fully understand indirect dependency exposure.

A vulnerable downstream package can create exposure even when organizations never intentionally install compromised software directly.

Visibility matters.

Inventory matters.

Credential isolation matters.

This incident reinforces a growing reality.

Software supply chain security is no longer optional.

It is foundational infrastructure defense.

Fact Checker Results

✅ Microsoft reported an active npm supply chain attack involving compromised @antv packages.

✅ The malware specifically targeted CI/CD secrets, GitHub Actions environments, and cloud credentials.

❌ The attack is not ongoing through currently remediated @antv packages, as maintainers and ecosystem defenders have already taken corrective actions.

Prediction

🔮 Software supply chain attacks will continue increasing as attackers pursue scalable compromise methods.

🔮 Development pipelines will receive stronger built-in security protections focused on runtime secret isolation and dependency validation.

🔮 Organizations will increasingly treat CI/CD infrastructure as critical production infrastructure rather than development tooling.

🕵️‍📝Let’s dive deep and fact‑check.

References:

Reported By: www.microsoft.com
Extra Source Hub (Possible Sources for article):
https://www.stackexchange.com
Wikipedia
OpenAi & Undercode AI

Image Source:

Unsplash
Undercode AI DI v2
Bing

🔐JOIN OUR CYBER WORLD [ CVE News • HackMonitor • UndercodeNews ]

💬 Whatsapp | 💬 Telegram

📢 Follow UndercodeNews & Stay Tuned:

𝕏 formerly Twitter 🐦 | @ Threads | 🔗 Linkedin | 🦋BlueSky | 🐘Mastodon | 📺Youtube