malicious-packages database. There is no coordinated-disclosure window (the maintainer is the attacker). codelake never stores raw payloads — only the detection rule, file:line, and IOCs.
matrix-by-lmx — malicious npm package, independently detected and OSV-corroborated
matrix-by-lmx is a malicious npm package that codelake Research independently detected during automated supply-chain scanning, corroborated by MAL-2026-15670 in OSV. The detection signal and affected versions are detailed below.
codelake's analysis classifies the detection signal as behaviour that exfiltrates local data to an attacker-controlled sink; harvests local credentials, SSH keys, cloud secrets or API keys.
Affected version(s): 1.4.13, 1.4.2, 1.4.3, 1.4.4, 1.4.9.
codelake's per-ecosystem scanners flagged the specimen and the AI-triage judge classified it malicious (recorded signal: live GCP API key exfil, confidence 0.90).
This is an independent detection by codelake's own scanners and AI-triage pipeline; it is corroborated by MAL-2026-15670 in OSV (codelake was not necessarily the first public reporter for this record).
All analysis is performed at the indicator level on extracted source only — codelake never executes specimens and stores only the detection rule, indicators and analysis, never raw payloads.
Detected independently by codelake Research · automated per-ecosystem supply-chain scanning + AI-triage, verified at the indicator level on extracted source (no execution). Corroborated by MAL-2026-15670 in OSV. codelake stores only the detection rule, indicators and analysis, never raw payloads.