plumery

AI Security Review

Evidence context: threat category credential theft; evidence quality strong.

This package presents an extreme security risk and should be treated as an active threat. Unlike typical SDKs that might trigger noise from IoC scanners or code-smell rules, 'plumery' exhibits indicators of confirmed malicious activity: it has 8 critical secret findings and 545 high-severity malware signature hits. While legitimate bundles often flag numerous IoCs (domains, IPs) and code-smell findings (e.g., 'eval', 'env access'), they do not trigger high-severity malware signatures. The presence of 545 malware signatures, combined with 8 critical secrets, strongly suggests this is either a trojanized dependency designed to harvest and exfiltrate credentials or a known malware family being distributed under the guise of a legitimate API library. The 'official' naming and 'unknown' version further align with typosquatting or brand impersonation tactics. There is no legitimate reason for a standard TypeScript API library to trigger this volume of high-severity malware heuristics.

Key Reasons

• 545 high-severity malware signature hits are definitive indicators of malicious code/payloads
• 8 critical secret findings (credentials) suggest active harvesting or hardcoded backdoors
• Extreme risk score (100) and trust score (0) correlate with confirmed threat patterns
• Metadata (unknown version, 'official' description) is consistent with typosquatting/supply chain attacks
• Tool-poisoning count is zero, ruling out standard library/SDK false positive patterns

False Positive Considerations

• High volume of IoCs (1515) are likely bundled dependencies in dist/
• Medium severity findings are likely generic code patterns

Reviewed 2026-04-12; recommended action: takedown request; model confidence 92%.