A npm package copying the official ‘postmark-mcp’ project on GitHub turned bad with the latest update that added a single line of code to exfiltrate all its users' email communication.
Published by a legitimate-looking developer, the malicious package was a perfect replica of the authentic one in terms of code and description, appearing as an official port on npm for 15 iterations.
Model Context Protocol (MCP) is an open standard that allows AI assistants to interface with external tools, APIs, and databases in a structured, predefined, and secure manner.
Postmark is an email delivery platform, and Postmark MCP is the MCP server that exposes Postmark’s functionality to AI assistants, letting them send emails on behalf of the user or app.
As discovered by Koi Security researchers, the malicious package on npm was clean in all versions through1.0.15, but in the 1.0.16 release, it added a line that forwarded all user emails to an external address at giftshop[.]club linked to the same developer.
Dev added their email address to receive copies of users' communication
Source: Koi Security
This extremely risky functionality may have exposed personal sensitive communications, password reset requests, two-factor authentication codes, financial information, and even customer details.
The malicious version on npm was available for a week and recorded around 1,500 downloads. By Koi Security's estimations, the fake package might have exfiltrated thousands of emails from unsuspecting users.
For those who downloaded postmark-mcp from npm, it is recommended to remove it immediately and rotate any potentially exposed credentials. Also, audit all MCP servers in use and monitor them for suspicious activity.
... continue reading