The Agent Skills Directory

Remote Code Execution (CRITICAL): The skill instructs the execution of curl -fsSL https://cli.inference.sh | sh. This is a piped remote execution pattern from an untrusted source (inference.sh is not in the trusted organization list).
Evidence: Found in SKILL.md under the 'Quick Start' section.
Risk: This allows the remote server to execute arbitrary code on the user's machine with current user privileges.
External Downloads (HIGH): The skill promotes the installation of additional unverified skills using npx skills add inference-sh/skills@....
Evidence: Found in the 'Related Skills' section of SKILL.md.
Risk: Dynamically pulling code from unvetted repositories increases the attack surface for supply chain attacks.
Indirect Prompt Injection (LOW): The skill acts as a template for generating video prompts which are then interpolated into tool calls.
Ingestion points: User-provided subjects, actions, and styles in SKILL.md.
Boundary markers: Absent. Prompts are directly interpolated into JSON strings for the infsh command.
Capability inventory: Access to the Bash tool with infsh command capability.
Sanitization: None detected. Malicious user input could potentially alter the command structure or parameters sent to the infsh tool.

video-prompting-guide