GitHub attack hits 61,000 repositories, steals cloud and AI credentials
Attackers took over 4,837 GitHub accounts and used them to target 61,764 repositories. At least 1,835 repositories executed the malicious code, leaking cloud credentials for AWS, Azure, Google Cloud, and AI services like OpenAI and Anthropic.
Editor, Lazyfounder
30 SEC SUMMARY
- Attackers took over 4,837 GitHub accounts and used them to target 61,764 repositories, turning GitHub Actions into a backdoor for malicious code.
- A 1,500-line script hidden in pull requests stole credentials for AWS, Azure, Google Cloud, OpenAI, and Anthropic from at least 1,835 repositories.
- The malware altered cloud settings to keep access even after victims reset their credentials, then wiped logs in under a minute to hide its tracks.
- Compromised PHP packages on Packagist and two Go modules showed the attack had spread beyond GitHub.
TABLE OF CONTENTS
The scale of the GitHub attack
Cybersecurity firm Aikido uncovered an attack that turned 4,837 stolen GitHub accounts into a launching pad for malicious code. Those accounts targeted 61,764 repositories, and at least 1,835 of them ran the attackers’ scripts, according to TechRadar.
How attackers breached the repositories
Attackers used stolen GitHub credentials to hijack GitHub Actions, a tool designed to automate tasks like pull requests. They opened pull requests titled “docs: dependency audit notes” from branches named “wfx-docs” to test which repositories they could breach.
Once inside, they embedded malicious scripts in three ways: editing existing workflows, attaching a shell script to npm install commands, or creating a new workflow file on a temporary branch.
What the malware stole and how it hid its tracks
The attack relied on a 1,500-line script that scoured infected systems for secrets. It stole GitHub Actions tokens, AWS, Azure, and Google Cloud credentials, Docker and Kubernetes configurations, SSH keys, and API keys for OpenAI and Anthropic. It also grabbed environment files, private keys, and credentials for npm, PyPI, HashiCorp Vault, and Terraform.
After stealing the data, the malware tried to expand its reach. It exchanged tokens for cloud sessions, created fake service accounts, and generated new credentials. It even altered AWS and Azure settings to keep access after victims reset their passwords. Within a minute, the attackers reverted their changes and deleted workflow logs to erase evidence.
Attack spreads beyond GitHub
The attack didn’t stop at GitHub. Compromised PHP packages appeared on Packagist, and two Go module versions were also affected. Aikido’s researchers believe the attackers likely triggered automated publishing pipelines rather than directly targeting these ecosystems.
Immediate steps for affected users
Aikido warns any repository owner whose project ran the malicious script to treat every secret and cloud credential as compromised. Immediate rotation of all passwords, API keys, and tokens is critical to block further unauthorized access.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
When developers store cloud credentials in GitHub workflows without strict rotation rules, a single stolen password can give attackers a foothold to drain secrets from thousands of projects in hours.
Key takeaways
- Owners of any repository that ran the malicious script must rotate every secret and cloud credential immediately—no exceptions.
- Teams should disable auto-merging for pull requests from unfamiliar branches like 'wfx-docs' and require manual review.
- Cloud providers could add real-time alerts when GitHub Actions identities modify AWS or Azure configurations.
Related on Lazyfounder
Sources
- TechRadar · 2026-11-13
Thousands of GitHub repositories targeted in widespread attack — 4,800 accounts already compromised to steal cloud credentials
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.
