Cybersecurity researchers have found a set of trojanized npm packages that masquerade as working calendar and streak utilities however are engineered to stealthily ship a synthetic intelligence (AI)-powered Linux implant dubbed RedC2 4.0.
“When the module masses, it locates the bundled binary, marks it executable, and launches it as a indifferent background course of,” TrendAI, Development Micro’s enterprise cybersecurity enterprise, mentioned in a report revealed Thursday. “No set up hook perform name is required; a single import wherever within the dependency graph, even a transitive one, is sufficient to execute the payload.”
The record of recognized packages is beneath –
- streak-metrics-math@1.0.0,1.0.1
- kit-map-vim@1.0.0
- streak-map-cache@1.0.0
- streak-map-kit@1.0.0
- map-streak-kit@1.0.0
- streak-cache-map@1.0.0
- streak-calc-metrics@1.0.0
- streak-calc-math@1.0.0
- streak-math-abz@1.0.0
- streak-metricsaz@1.0.0
- streak-math-metrics@1.0.0
- streak-metricazbd@1.0.0
- streak-metricsazb@1.0.0
- streak-kit-map@1.0.0
What’s notable about these packages is that they’re practical and supply the promised performance. However beneath that garb of date utilities is code designed to drop a Linux backdoor by framing it as a local math accelerator. The title of the file varies throughout the packages: math-core.bin, math-calc.bin, calc-math.dat, calc-cache.bin, calc.bin, calc-mapping.bin.
It is situated both straight inside the “dist/” or below “dist/inside/,”, however what it incorporates is identical: the RedShell Linux beacon for RedC2 4.0 that communicates with a distant Home windows or Linux server to facilitate post-exploitation actions on the compromised host.
“Supply is dealt with by the bundle entry file, dist/index.mjs, which acts as a trojan loader,” safety researcher Aliakbar Zahravi mentioned. ”It re-exports the date helpers and launches the bundled implant as quickly because the module masses, with no set up hook and no exported perform required.”
RedC2 4.0, marketed on cybercrime boards as a cross-platform toolkit for Home windows, macOS, and Linux, gives surveillance, credential theft, payload loading, and mass-operation capabilities. The model was marketed by a risk actor named “MarlboroMan” on Hack Boards in early June 2026, describing it as a command-and-control (C2 or C&C) framework “constructed for evasion.”
Model 3.0 of RedC2 was offered earlier this January, whereas model 2.0 was launched in August 2025, indicating the framework has been below lively improvement for at the least a yr. The RedShell Linux beacon was launched in model 4.0.
The C2 framework can be feature-rich, supporting terminal entry, file switch, staged payload supply, knowledge assortment, multi-beacon operation, community visualization, host-to-host tunneling, and in-memory execution of Beacon Object Information (BOFs), .NET assemblies, and shellcode.
 |
| RedShell Linux execution stream |
The Linux variant of the beacon, as soon as deployed, gives an interactive shell via “/bin/sh” and exposes Linux-specific instructions to allow system discovery, file operations, knowledge assortment (e.g., SSH keys and browser credentials), execution, persistence, in-memory ELF execution, SOCKS5 proxying, and community pivoting.
It additionally establishes communication with a C2 server and registers the contaminated system by gathering primary system info and transmitting it within the type of a “check-in message,” after which it enters a command-processing loop to course of incoming directions from the operator, execute them through “/bin/sh,” and ship the outcomes again.
The Home windows and macOS counterparts cowl an analogous floor, permitting file operations, host and community reconnaissance, consumer enumeration, and knowledge harvesting. The Home windows beacon additionally incorporates Consumer Account Management (UAC) bypass, antivirus and endpoint detection, antivirus tampering, in-memory execution, and lateral motion that the macOS model lacks.
On a clearnet web site branded Purple Offsec, the risk actor claims, “Purple C2 is a multi-language, multi-OS command and management framework designed for Home windows, Linux, and macOS. Your complete framework was constructed with evasion as a core precept, using the most recent developments and strategies within the offensive safety area.” It is obtainable for buy for $99.99.
Purple Offsec’s Phrases of Service expressly prohibit its clients from utilizing the instrument for “unauthorized pc entry,” “hacking with out express permission,” and “abuse, exploitation, or injury of techniques you don’t personal or usually are not licensed to check.”
“Purple Offsec gives instruments meant for purple crew professionals and customers who perceive exterior offensive safety tooling inside authorized and moral boundaries,” the phrases learn.
RedC2 extends its management layer with a command-line extension known as RedC2 EXT in addition to a big language mannequin (LLM)-driven part known as Purple Agent, the latter of which lets operators orchestrate advanced post-exploitation duties, similar to community reconnaissance and credential dumping, utilizing pure language instructions.
“RedC2 ships with an AI assistant known as Purple Agent, an LLM-backed command execution layer that turns natural-language intent into framework beacon instructions,” Zahravi mentioned. Purple Offsec has characterised it as an “AI-powered command execution system specialised for penetration testing.”
The findings underscore how beforehand undocumented AI-integrated C2 frameworks are being distributed through malicious npm packages, whereas concurrently decreasing the barrier to entry.
“By interacting with a mannequin tuned for red-team operations, an operator inputs natural-language prompts, and the framework interprets them into actionable command sequences,” TrendAI mentioned. “This abstraction lets operators of various ability ranges execute advanced, multi-stage intrusions effectively.”
The event comes shut on the heels of a coordinated provide chain assault affecting three official Rust crates (arrayref@0.3.10, internment@0.8.7, and append-only-vec@0.1.9), compromising them with a malicious proc-macro1 dependency that executed cross-platform malware routinely throughout Cargo builds.
The malware is designed to profile the contaminated machine, catalog Chromium-based browsers, set up persistence, and beacon to attacker-controlled infrastructure for tasking and downloading extra payloads.
It is suspected that the maintainer’s publishing credentials had been compromised to push the poisoned variations to the bundle repository. Proof factors to infrastructure overlaps with prior software program provide chain assaults concentrating on Mastra and Axios, each linked to North Korean risk actors.
