The @ch4acko3/frontal-lobe npm package contains embedded malicious functionality that exfiltrates sensitive AI session data to a hardcoded IP address (http://47.112.15.137:8789/frontal/trace-sync) over unencrypted HTTP. The postinstall lifecycle script automatically enables this data transmission by writing a configuration file with tracing enabled, prompts the user for a user ID, and silently reuses it for identity in the data uploads. The data sent includes gzip+base64 encoded chunks of user prompts, model outputs, source code context, and trace events containing memory query and recall text. This exfiltration occurs without a distinct opt-in or user consent, and the endpoint is not configurable, differing from the documented service endpoint. This behavior constitutes a malicious package designed to leak sensitive user data.

Sensitive user data including AI session prompts, model outputs, source code context, and memory trace events are exfiltrated to an unauthorized external server over unencrypted HTTP. This can lead to confidentiality breaches of proprietary or private information. The data exfiltration is automatic upon installation without explicit user consent, increasing the risk of unnoticed data leakage.

No official patch or remediation is currently documented. Users should avoid installing or using the affected versions (=0.1.12, =0.1.12-preview.2, =0.1.14) of the @ch4acko3/frontal-lobe package. Review and remove the package if already installed. Monitor for any unusual network traffic to the hardcoded IP address (47.112.15.137) and consider network-level blocking of this endpoint. Patch status is not yet confirmed — check the vendor advisory or trusted sources for updates.