Malicious code in @marketfront/madvpopup (npm)
The @marketfront/madvpopup package is part of a 25-package malicious campaign batch-published to the @marketfront npm scope by npm user 'marketfront' ([email protected]) within a roughly 3-minute window on 2026-07-01. All packages in the campaign were published at version 7.0.0 and use e-commerce/marketing frontend component names as cover.
The package declares a postinstall hook (node scripts/postinstall.js) that executes heavily obfuscated (obfuscator.io-style) code automatically at npm install time. Static analysis of the decoded payload revealed a credential harvester that dynamically requires fs, os, http, https, zlib, path and dns, then reads approximately 20 sensitive credential files including ~/.ssh, ~/.aws/credentials, ~/.kube/config, ~/.docker/config.json, ~/.npmrc, ~/.netrc, ~/.pgpass, ~/.git-credentials, ~/.env and ~/.bash_history. Collected data is exfiltrated via a gzip-compressed HTTPS POST with a custom X-Secret header to the path /api/v1/events, alongside a DNS resolver beacon. The command-and-control host is concealed behind an additional RC4+XOR encryption layer around an embedded configuration blob and was not statically resolved.
The decoded behavioral payload (module requires, credential-file target list, exfiltration headers and endpoint) is byte-for-byte identical across sampled packages in the campaign. The campaign shares tooling and infrastructure patterns (obfuscated postinstall credential harvester, X-Secret header, /api/v1/events exfiltration path, RC4-concealed C2) with the earlier @emcd-vue campaign, indicating the same actor rotating scopes and disposable maintainer emails.
Source: amazon-inspector
Package declares postinstall: node scripts/postinstall.js. scripts/postinstall.js is a 162KB obfuscator.io-style payload (string-array + RC4 + base64 decoders, per-call scope aliases, and anti-debug: NODE_OPTIONS inspection, inspector-module check, Date.now() tight-loop timing probe gating execution via a global flag). On npm install the script collects hostname, network interfaces, cwd, uid/gid, home, tmpdir, uptime, username, npm_config_user_agent, Windows-specific variables (USERDOMAIN, COMPUTERNAME, PROCESSOR_ARCHITECTURE, APPDATA, LOCALAPPDATA, PROGRAMDATA, TEMP), and the full process.env, and POSTs them over HTTPS to a hardcoded (obfuscated) remote endpoint. A dedicated helper iterates process.env against a large embedded allowlist of credential-shaped variable names (AWS/DB/CI/token patterns) and includes matches in the outbound payload — direct credential harvest against installer machines and CI runners where cloud/registry tokens live in the environment. The library surface is non-functional cover: dist/index.js re-exports ../src/index.js, which is not shipped in the tarball; the only executable code is the postinstall. Scope and metadata (@marketfront/, homepage docs.marketfront.io, repo github.marketfront.io, bugs jira.marketfront.io, README instructing .npmrc redirect to npm.marketfront.io) impersonate an internal-corporate namespace, consistent with a dependency-confusion attack against a specific organization's developers.
