The colorpicker-ui npm package includes a postinstall hook that executes a setup.js script during installation. This script reads sensitive credentials such as Kubernetes service-account tokens from /var/run/secrets/kubernetes.io/serviceaccount/token, AWS instance metadata from the IMDS endpoint, and local system files like /etc/passwd, /etc/resolv.conf, and /etc/hosts. It collects environment variables and system information (id, uname -a), recursively lists the /app directory, and writes the harvested data into files named _rce.ts and rce.json placed in paths intended to be imported by AI coding-agent sandbox workspaces. This mechanism enables arbitrary code execution in subsequent agent runs and exfiltrates secrets to attacker-controlled processes. The legitimate color-picker API is functional but serves only as a facade to conceal the malicious payload.

Compromise of Kubernetes cluster credentials, cloud instance credentials, and local system information can lead to unauthorized access, lateral movement, and persistent control over affected environments. The malicious files created enable arbitrary code execution within AI coding-agent sandbox environments, increasing the risk of further compromise and data exfiltration. This threat affects any environment where the vulnerable package versions are installed, potentially exposing sensitive infrastructure and application secrets.

No official patch or remediation is currently documented. Users should immediately avoid installing or remove versions 1.2.4, 1.2.5, and 1.2.6 of colorpicker-ui. Audit existing environments for the presence of the malicious files (_rce.ts, rce.json) and any suspicious postinstall activity. Consider using trusted package sources and verifying package integrity before installation. Monitor for any unexpected code execution within AI coding-agent or similar sandbox environments. Patch status is not yet confirmed — check the vendor advisory or npm security advisories for updates.