The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requests broad permission to execute the infsh CLI tool via allowed-tools: Bash(infsh *). This tool is the primary interface for interacting with AI models on the inference.sh platform as described in the documentation.
[EXTERNAL_DOWNLOADS]: The skill references image assets hosted on the vendor's official domain (cloud.inference.sh) and suggests installing related skills from the same vendor using npx. These are legitimate vendor-originated resources.
[SAFE]: No malicious patterns such as credential exfiltration, obfuscation, or unauthorized persistence were detected. The inclusion of potentially vulnerable code examples is for educational purposes within the context of security-focused prompt engineering.

prompt-engineering