Investigate credential theft and data exfiltration, including attempts to send secrets or other sensitive data to an outside recipient. These reports include LPM Firewall detections. Each report identifies the package version, supporting evidence, advisory sources, and current Firewall decision.
Updated 9 Oct 2026, 21:10 UTC. Ordered by latest scan.
This is concrete, repeated exfiltration of local AI-agent conversations triggered by routine use, with no README disclosure or per-launch consent. It is not an install-hook case, but is m...
Direct source inspection confirms an automatic install-time secret-harvesting and exfiltration chain. This is concrete malicious behavior, not a package-aligned network use.
Direct source inspection confirms an automatic install-time credential and wallet-file exfiltration chain. This is concrete malicious behavior, not a noisy static signal.
Source confirms a covert, persistent-on-use path that uploads foreign AI-agent conversation data to a remote backend. The absence of install hooks does not mitigate this concrete unprompt...