The Miasma credential theft assault framework has lately focused the open supply ecosystem by means of provide chain assaults and was briefly open sourced on GitHub.
Miasma seems to be an evolution of the sooner Shai-Hulud worm that was beforehand leaked on GitHub, sharing lots of the identical options, methods, and even code.
The malware infects developer machines, steals construct atmosphere and cloud credentials, makes use of them to compromise legit repositories and packages, publishes trojanized variations to contaminate downstream builders, and repeats the cycle.
This autonomous, worm-like self-propagation mechanism can quickly broaden its attain, turning a single breach right into a widespread provide chain assault.
This malware has beforehand been related to high-profile assaults in opposition to Pink Hat npm packages, and extra lately with assaults in opposition to 73 Microsoft repositories on GitHub.
SafeDep researchers reported yesterday that Miasma’s supply code was leaked to GitHub through a lot of compromised developer accounts. In every of those accounts, the attackers leaked supply code in a repository named “Miasma-Open-Supply-Launch.”
This means that the attacker deliberately launched the supply code, somewhat than an unintentional leak, much like the earlier launch of Shai-Hulud’s code.

Supply: Safedep
Evaluation of the code revealed that the toolkit doesn’t require command and management (C2) infrastructure to function, because it makes use of GitHub for its goal.

The framework collects credentials from cloud suppliers, CI/CD programs, password managers, Kubernetes, and secret shops and exploits them to compromise npm, PyPI, and RubyGems packages, in addition to GitHub repositories, Actions workflows, and JFrog Artifactory cases.
They will additionally transfer laterally by means of SSH and AWS Techniques Supervisor (SSM) to infect the configurations of AI coding instruments similar to Claude, Gemini, Cursor, Copilot, Kiro, and Cline.

Supply: Safedep
One fascinating function revealed within the leaked Miasma supply code is a “lifeless man change” that’s put in when the malware makes use of a sufferer’s stolen GitHub tokens as an exfiltration channel.
This part screens the validity of the token each minute and executes a harmful command if the token is revoked (rm -rf ~/; rm -rf ~/Paperwork), recursively deletes information and directories within the consumer’s house and paperwork folders.
The monitor runs as a systemd consumer service on Linux and as a LaunchAgent on macOS and stays lively for as much as 72 hours.
One other fascinating facet revealed is the five-stage construct pipeline that generates a novel payload for every construct.
SafeDep stories that this course of combines per-file AES-256-GCM encryption of embedded property, randomized string obfuscation, supply transformation, JavaScript obfuscation, and a self-extracting loader that wraps the ultimate payload in three layers of encryption.
The random key and randomized exterior encoding layer make sure that every generated pattern is completely different from earlier builds, making signature-based detection and static evaluation troublesome.
The Shai Hulud leak led to the discharge of extra superior variants similar to Miasma, which elevated assault charges. Equally, the leak of Miasma’s supply code is predicted to have an analogous influence as menace actors undertake and additional tweak the code.
This could have a big influence on the safety of open supply ecosystems, as provide chain assaults proceed to focus on them at an unprecedented tempo.
Software program builders are inspired to lock down undertaking dependencies, introduce a multi-day delay earlier than adopting newly launched package deal updates, and validate new builds in an remoted take a look at atmosphere.
Safety groups doc 54% of profitable assaults and situation a warning on solely 14%. The remainder strikes invisibly by means of the atmosphere.
Picus’ whitepaper exhibits find out how to take a look at your SIEM and EDR guidelines in breach and assault simulations to make sure threats go undetected.
Get the white paper
