attacking-serverless

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This document contains explicit, actionable instructions and code examples for stealing credentials, exfiltrating environment data, installing persistent backdoors, and abusing IAM to escalate privileges — indicative of deliberate malicious activity.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly instructs persistence and modification of the execution environment (writing a backdoor to /tmp, reading /proc/self/environ) and guides creating/replacing functions and attaching IAM policies — actions that alter the running machine or its cloud resources and thus compromise host state.