Nearly 9,000 GitHub Repositories Hijacked by Stealthy wfx Supply Chain Attack
Because nothing says 'cutting-edge AI development' quite like letting a malicious script quietly pack up your cloud keys and ship them off to strangers in the middle of the night.
A routine dependency update turned into a digital break-in when a compromised version of the NeuronAI Studio package dropped silently into the Packagist registry. Designed to help developers stitch together visual AI agents on Laravel, this particular update came bundled with a hidden passenger known to researchers as the wfx credential stealer. As soon as automated build tools fetched the package, the script immediately went to work mapping out the environment and hunting down sensitive access tokens.
The attackers didn't smash doors down; they politely knocked, let themselves in, and cleaned out the safe while pretending to be helpful maintenance bots. The operation relied entirely on stolen GitHub credentials to slip past the perimeter. First came an innocent-looking pull request tweaking some documentation just to test write permissions, followed swiftly by sneaky automated commits disguised as routine build cache refreshes. Once the script harvested every cloud key and publishing token within reach, it meticulously erased its own footprint by rolling back the commit and wiping the workflow run logs clean off the dashboard.
Security teams quickly scrambled to patch the bleeding, though initial fix attempts stumbled when subsequent updates also landed on malware blacklist registers alongside older builds like versions 3.14.2 and 3.14.3. The infection didn't stop at simple repository snooping either, as the script aggressively poked at cloud infrastructure by tweaking trust policies for AWS IAM roles and injecting rogue credentials into Azure environments. Similar malicious payloads soon turned up hiding inside unrelated open-source modules across other language ecosystems, proving that trusting automated build pipelines is basically playing Russian roulette with a fully loaded CI server.
Supply chain security remains a hilarious exercise in hoping that nobody slipped rat poison into the open-source smoothie everyone drinks for breakfast every single day.
Source: Endor Labs
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.