The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the infsh CLI to interact with the inference.sh platform. It executes commands for authentication (infsh login), listing available apps (infsh app list), and running specific AI models (infsh app run). These operations are the primary function of the skill and are performed within the context of the user's CLI environment.
[EXTERNAL_DOWNLOADS]: The skill references the installation of the inference-sh/skills package via npx to obtain the necessary CLI tools. It also handles various media assets by downloading them from external URLs (e.g., cloud.inference.sh, falai) during the video generation process, which is standard for cloud-based AI inference.
[PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection surface as it interpolates user-provided text (e.g., prompts for TTS) and media URLs directly into CLI command arguments without explicit sanitization or boundary delimiters. While this is a vulnerability surface, it is considered acceptable within the context of the skill's intended use for media generation.
Ingestion points: Input parameters such as image_url, audio_url, prompt, and text are extracted from user or data context in SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the command templates.
Capability inventory: The skill has access to bash command execution via the infsh tool across all examples in SKILL.md.
Sanitization: No input validation or character escaping is specified for the external data being processed.

ai-avatar-video