{"id":"MAL-2026-13880","revision":"2b853ae1395143739d898c6761a4bff0","records":[{"revision":"2b853ae1395143739d898c6761a4bff0","observedAt":"2026-10-24T13:17:17.793+00:00","statement":{"source":"osv","id":"MAL-2026-13880","aliases":[],"title":"Malicious code in mcp-util-helpers (npm)","description":"\n---\n_-= Per source details. Do not edit below this line.=-_\n\n## Source: amazon-inspector (2044da43e53933094b69c1436146bdea64cbc4c011e4b8a141ceea0cab19eb5c)\nOn `npm install`, the preinstall hook (scripts/check-env.js) runs a host-reconnaissance sweep — whoami, id, uname, ip/route, ss listeners and connections, ps, docker ps, crontab, /etc/passwd, /etc/hosts, environment variables filtered for key/token/pass/secret/api/auth, and a listing of ~/.ssh/ — and POSTs the collected data via curl to a hardcoded webhook.site bucket at https://webhook.site/0c5803fc-bdd6-4dd0-8c57-68ee0cabac29/the analysis. The script then parses the webhook response as JSON: if it contains a `cmd` field the value is passed to shell exec; if it contains a `url` field the referenced file is downloaded to /tmp/.mcp-util-setup, chmod +x'd, and executed, with the output POSTed back to the same bucket at /the assessment. This is install-time credential/host reconnaissance combined with an operator-controlled remote code execution channel.\n","state":0,"publishedAt":"2026-08-12T12:39:48+00:00","modifiedAt":"2026-08-12T13:00:23.4467407+00:00","affected":[{"ecosystem":"npm","name":"mcp-util-helpers","ranges":[],"versions":["1.0.0"],"rawVersionEvidence":null}],"metrics":[],"references":["https://www.npmjs.com/package/mcp-util-helpers/v/1.0.0"],"sourceUrl":"https://osv.dev/vulnerability/MAL-2026-13880","license":"OSV · upstream source license applies","key":"osv:MAL-2026-13880","identifiers":["MAL-2026-13880"]},"retained":false,"historicalAffected":[]}],"coverage":[]}