ssti

SUSPICIOUS. The skill is internally coherent and has low supply-chain and exfiltration risk, but it is a security/exploit-oriented agent skill that teaches SSTI discovery and validation with RCE payloads. Its footprint matches its stated purpose, yet enabling offensive security testing for an AI agent makes it high-risk despite no signs of malware or credential theft.

The provided reference highlights a pervasive risk: many popular templating engines can enable remote code execution when used with untrusted input, depending on version and configuration. To mitigate risk, enforce sandboxing, pin safe versions, validate and constrain input, and prefer engines with proven isolation or minimal capabilities for user-controlled templates. This content serves as a risk awareness guide rather than executable code, with actionable guidance limited to best-practice mitigations.

This fragment is explicit SSTI-to-RCE exploit/PoC material. It demonstrates how attacker-controlled template strings can be compiled/rendered directly to reach runtime internals (Python os and Node child_process) and execute a system command ('id'). While it is framed as a skeleton rather than proven integration into a real library API, its content is unambiguously weaponized and would pose a critical security threat if distributed in or executed as part of production logic that renders untrusted templates.