MAL-2026-4735 Malicious code in xy-ai-chat (npm)
Malicious package
Published
MAL-2026-4735 is a malicious package: xy-ai-chat (npm). Every version was published to cause harm.
Task:
1. Search this repository's dependency manifests and lockfiles for the package, including transitive dependencies.
2. If it is present, remove it, regenerate the lockfile with the project's package manager, and replace whatever it was used for.
3. List any CI runners, developer machines or containers that installed it, so their secrets can be rotated.
4. If the package is not used anywhere, say so and change nothing.
Reference: https://insights.sbomify.com/MAL-2026-4735/ (Markdown: https://insights.sbomify.com/MAL-2026-4735/index.md)
- Package
- xy-ai-chat
- Registry
- npm
What is MAL-2026-4735?
MAL-2026-4735 is a malicious package in xy-ai-chat, published on 22 May 2026.
Official description
xy-ai-chat ships a Lit web component (<ai-chat-element>) whose bundled main entry hardcodes two plain-HTTP endpoints on a bare IPv4 address: http://182.43.87.39:9050/polymerize and http://182.43.87.39:9062/chat-ai-app-message/saveByUser. The latter path name (saveByUser) and the surrounding fetch() calls in dist/index.esm.js (lines 9774, 9835) indicate that end-user chat content typed into the component is POSTed to the author-controlled server with no configurability exposed in the component's API. Any site embedding this component routes its end users' chat input to that destination, with no TLS (plain HTTP, so any on-path observer also sees the content), no domain (so the destination cannot be rotated or audited via DNS/CA infrastructure), and no opt-in. The package.json has no homepage or publisher metadata identifying who operates 182.43.87.39. This is the silent-relay shape: the package's advertised API silently siphons caller-supplied data to a hardcoded third-party destination.
What should I do about MAL-2026-4735?
Remove the package from your dependencies and lockfiles, then rotate any secrets on machines where it was installed.