Fake Web3 Accelerator Uses npm Packages to Spread Malware and Hijack AI Agent Traffic

Executive Summary

A fake accelerator used malicious npm packages to compromise developer systems, steal information, and interfere with AI-agent and LLM traffic. Organizations running coding agents, MCP integrations, or third-party AI tooling should prioritize dependency and agent-configuration controls.

What Happened

A fake Web3 accelerator has been observed pushing information stealers and backdoors to Windows and macOS systems using malicious npm packages under the pretext of offering free artificial intelligence (AI) credits for Anthropic Claude and Google Gemini, coding tools, and security audits. "The tools delivered information-stealing malware and backdoors for macOS and Windows machines," Elastic

Affected Products

Anthropic Claude
Google Gemini
OpenRouter
npm
@apexacc/cli
@apexacc/audit
MCP server
Apex Copilot CLI

Vulnerability / Attack Technique

Malicious npm package and AI agent traffic interception

Technical Mechanism

The malicious npm packages were distributed under the guise of AI credits, coding tools, and security audits, but installed an information stealer or backdoor on Windows and macOS systems. The packages could intercept and redirect OpenRouter LLM traffic and register an attacker-controlled MCP server in the victim's agent configuration.

AI Security Category

LLM03:2025 Supply Chain

Part of the MCP Security topic.

Part of the AI Supply Chain and AI SBOM topic.

Relevant Frameworks

LLM03:2025 Supply ChainMITRE ATLAS: Initial AccessMITRE ATLAS: ExecutionMITRE ATLAS: PersistenceMITRE ATLAS: CollectionMITRE ATLAS: Exfiltration

RealGround Analysis

This incident shows how compromised developer dependencies can affect both host security and the integrity of AI-agent workflows. An attacker-controlled package or MCP server may redirect model traffic, influence tool interactions, and expose sensitive agent context, making dependency provenance and runtime agent visibility essential for teams operating LLM applications.

Recommended Controls

Require allowlisted npm packages, publisher verification, lockfiles, and mandatory malware scanning before installation.
Maintain an SBOM covering AI agents, npm dependencies, MCP servers, model providers, and LLM-routing services.
Restrict agent configuration changes and require review or approval before registering MCP servers or changing model endpoints.
Monitor outbound model traffic for unexpected redirection, unauthorized endpoints, and sensitive data transmission.
Run AI red-team exercises that test malicious dependencies, compromised tools, unauthorized agent actions, and model-request interception.

Related RealGround Service

AI Agent SBOM and LLM Mapping · or start with the free AI Agent Security Scanner.

Sources

thehackernews.com: https://thehackernews.com/2026/10/fake-web3-accelerator-uses-npm-packages.html

Summary and analysis written by RealGround from the source above; last updated 2026-10-20.