Attackers use trusted GitHub repositories as malware staging infrastructure

Cybercriminals are increasingly abusing GitHub itself as a staging and delivery platform for malware, according to an analysis by penetration-testing firm Raxis, written by Brian Tant and republished on dev.to. The campaigns described there, including operations tracked by Microsoft and Trend Micro, have reached close to a million devices and rely on a simple asymmetry: attackers exploit the trust developers place in a platform they use every day.

Campaigns measured in hundreds of repositories

The largest case so far comes from Microsoft's analysis of Storm-0409, a malvertising campaign that infected close to one million devices worldwide. Separately, security researchers have identified more than 1,300 GitHub repositories vulnerable to RepoJacking, a technique that lets attackers take over existing projects and push malicious changes into code that developers already rely on. The Raxis post argues these are not opportunistic uploads but long-running operations built on a precise understanding of developer habits and supply-chain dependencies.

GitVenom: years of polish

The GitVenom campaign illustrates the patience involved. Over several years its operators stood up hundreds of repositories that looked legitimate: carefully written README files, plausibly generated with AI, realistic commit histories, and topics tuned to specific developer communities such as Instagram automation, Telegram bots, and game-cheating utilities.

Once downloaded, the code used obfuscated JavaScript or PowerShell to fetch additional payloads, including AsyncRAT and Quasar remote-access trojans and custom Node.js stealers. Some variants also hijacked the clipboard to redirect cryptocurrency transfers toward attacker-controlled wallets. Raxis cites evidence of roughly five bitcoin, around $440,000, moving to those wallets.

Malware delivered through GitHub Releases

Trend Micro uncovered a separate scheme centred on Lumma Stealer. Attackers created repositories with names resembling popular tools, then used GitHub's release feature to upload malware-laden executables. Because developers treat release downloads as official distributions, repository descriptions and release notes written in the style of genuine announcements completed the deception. The same channel distributed SectopRAT, Vidar, and Cobeacon alongside Lumma.

Storm-0409: from malvertising to NetSupport

Microsoft's Storm-0409 reporting describes a three-stage chain. Users of illegal streaming sites clicked malicious ads, were redirected to legitimate-looking GitHub repositories, and downloaded NetSupport remote-management tooling disguised as video codecs or media utilities. The operators kept multiple repositories running in parallel, creating redundancy that blunted takedown attempts.

Why GitHub suits attackers

Raxis attributes the platform's appeal to a mix of trust and convenient plumbing. A github.com URL reads as legitimate, both to humans and to the automated systems, including CI/CD pipelines, that fetch and execute code from repositories without extra vetting. Technically, GitHub's CDN delivers fast, reliable downloads worldwide, TLS encrypts all traffic and frustrates network-level inspection, Git's distributed model spreads code across mirrors and forks, and the platform's APIs let repository creation be automated at scale. Economically, free hosting removes the need for attackers to run their own distribution infrastructure and adds a layer of anonymity, while malware-as-a-service operators can rent payloads and delivery networks outright.

What teams can do

The post's recommendations are conventional but pointed: audit external dependencies and their sources on a regular schedule, wire security scanning into CI/CD pipelines, monitor which new repositories developers are pulling from, enforce clear code review and approval policies, and write incident-response procedures that specifically cover a compromised dependency or malicious repository.

Why it matters

The lesson is not that GitHub is broken; it is that a trusted domain says nothing about any particular artifact hosted on it. Nearly every professional codebase now pulls code, release binaries, and build steps from GitHub, and each touchpoint is a supply-chain dependency that can be forged. GitVenom also shows that AI assistance makes convincing fake projects cheap to produce at scale, so a professional-looking repository is no longer evidence of anything. Verification has to shift from where the code is hosted to what the code actually does, before it runs inside a pipeline or on a developer machine.