Browse / MCP and Tool Server Flaws / AML.CS0053
Poisoned Postmark MCP Server Email Exfiltration
- Also involves
- Unsafe Model Loading and Supply Chain, Tool and Agent Misuse, Misinformation and External Harm
- Target
- Postmark MCP Server (Incident)
How it worked
- Defense Evasion · Impersonation
The bad actor impersonated Postmark by publishing a legitimate version of their
postmark-mcppackage to npm. Postmark had not registered thepostmark-mcpname on npm themselves, allowing the bad actor to namesquat. Legitimate users were tricked into using the npm package even though it wasn't managed by the official developers ofpostmark-mcp - Resource Development · Develop Capabilities
The bad actor modified the legitimate Postmark MCP server to include their email address on the BCC line on all emails sent by the tool.
- Resource Development · Publish Poisoned AI Artifacts: AI Agent Tools
The bad actor published their malicious version of
postmark-mcpto npm. - Defense Evasion · AI Supply Chain Rug Pull
By waiting for users to adopt a legitimate version of
postmark-mcpfirst, the bad actor was able to evade the additional scrutiny and scanning performed on new tools. - Initial Access · Supply Chain Compromise: AI Agent Tool
When organizations upgraded
postmark-mcpto version1.0.16, they received the malicious version of the tool via the compromised supply chain. - Persistence · AI Agent Tool Poisoning: Implementation
Once configured with the organization's AI agents, the poisoned Postmark MCP server's effects persist.
- Execution · User Execution: Poisoned AI Agent Tool
When users at the victim organization instructed their AI agent to use tools provided by the poisoned Postmark MCP Server, the malicious code was executed.
- Exfiltration · Exfiltration via AI Agent Tool Invocation
When organizations sent emails via the
postmark-mcptool, the entire contents of their emails are exfiltrated to the bad actor via the address added on the BCC line. - Impact · External Harms
The exfiltrated emails may include transactional emails (revealing private information about the organization's clients) and promotional emails (revealing the organization's client list).
Summary
A bad actor successfully exfiltrated emails from users of the Postmark's MCP server via a supply chain attack. Postmark is an email delivery service that allows organizations to send marketing and transactional emails via API. The Postmark MCP server allows users to interact with Postmark via AI agents.
The bad actor impersonated Postmark, by registering the postmark-mcp package name on npm. They initially published the legitimate versions of the MCP server. After the package became popular and reached over 1,000 downloads per week, the bad actor performed a rugpull and uploaded a malicious version of the package. The malicious version added the bad actor's email address in the BCC line of all emails sent by the MCP tool. Users who upgraded to this version and continued to use the tool would have all emails exfiltrated to the bad actor.
References
- https://www.koi.ai/blog/postmark-mcp-npm-malicious-backdoor-email-theft
- https://postmarkapp.com/blog/information-regarding-malicious-postmark-mcp-package/
- https://snyk.io/blog/malicious-mcp-server-on-npm-postmark-mcp-harvests-emails/
Source: MITRE ATLAS case studies, used under Apache 2.0.