{"id":"MAL-2026-14482","revision":"75b096f87b5d43529d1ec41b2080b76d","records":[{"revision":"75b096f87b5d43529d1ec41b2080b76d","observedAt":"2026-10-24T13:17:17.793+00:00","statement":{"source":"osv","id":"MAL-2026-14482","aliases":[],"title":"Malicious code in vite-plugin-images-analysis (npm)","description":"\n---\n_-= Per source details. Do not edit below this line.=-_\n\n## Source: amazon-inspector (36ac1f55db3f6447b28ded9ee9c6b792487d225e52fe5ec393f75887001a5388)\nThis Vite plugin, advertised as an image-analysis tool, uses transformIndexHtml to inject an inline script into every built HTML page produced by the developer's application. The injected script reads end-user identity fields (account/username/userNo, phone, realName) from sessionStorage keys such as 'user' and 'user-store', hooks sessionStorage.setItem to capture values at login time, and transmits them via sendBeacon/fetch/Image requests. The destination URL is not present in cleartext in the source: the plugin's `dsn` option is a base64 string that is XOR-decoded with key 42 in the browser at runtime (atob(_c).split('').map(c=>String.fromCharCode(c.charCodeAt(0)^42)).join('')), and the outbound query string is likewise base64+XOR-encoded before being sent. The declared 'image analysis' purpose does not match the actual behavior of harvesting logged-in-user PII from the consuming application and posting it to a concealed endpoint.\n","state":0,"publishedAt":"2026-08-25T08:10:48+00:00","modifiedAt":"2026-08-25T23:50:02.9611949+00:00","affected":[{"ecosystem":"npm","name":"vite-plugin-images-analysis","ranges":[],"versions":["1.0.0"],"rawVersionEvidence":null}],"metrics":[],"references":["https://www.npmjs.com/package/vite-plugin-images-analysis/v/1.0.0"],"sourceUrl":"https://osv.dev/vulnerability/MAL-2026-14482","license":"OSV · upstream source license applies","key":"osv:MAL-2026-14482","identifiers":["MAL-2026-14482"]},"retained":false,"historicalAffected":[]}],"coverage":[]}