Radware researchers identified at least 83 malicious GitHub items with a verified payload across 61 public repositories in what they suspect is a coordinated campaign to execute commands in CI/CD environments and steal credentials and access tokens, according to The Register, which interviewed them for an exclusive published on 29 October 2026.

According to The Register, 77 of the items were published during a 43-minute span on 1 September and six more in the next two minutes; the attacker hid commands in HTML comments in issues, pull requests or comments so they were not visible on the rendered page, and in repositories configured to run an automated workflow on such events, a workflow that placed that content directly into shell source code ran the hidden command, which collects environment variables and attempts to send them to an attacker-controlled server.

The researchers told The Register they found two cases with evidence the command ran inside the automation environment but cannot say whether the attacker obtained repository write permissions, modified source code or compromised downstream software. Repositories listed include google/langextract, aquasecurity/trivy, mlflow/mlflow, brave/brave-browser, Homebrew/homebrew-cask and github/docs; the researchers say they suspect a single individual using 74 GitHub accounts and that they disclosed their findings to all the projects' maintainers; most of the repositories were closed or fixed, but the researchers cannot confirm that all of them were. (source: The Register)