The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions describe processing user-supplied data (JSON input files and text prompts) which are then passed to the runapi CLI tool for video generation. While this constitutes an attack surface for indirect injection, it is the primary intended function of the skill.
Ingestion points: Instructions mention using --input-file request.json and processing text/image prompts in SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the provided documentation.
Capability inventory: The skill utilizes the runapi CLI for execution and task management.
Sanitization: No specific sanitization or validation steps are defined in the instructions.
[EXTERNAL_DOWNLOADS]: The skill configuration and documentation reference several external resources for installation and integration, all originating from the vendor's official infrastructure.
The runapi CLI is installed via the runapi-ai/tap/runapi Homebrew formula.
Official SDKs are listed for multiple languages (Node.js, Python, Ruby, Go, Java, PHP) under the vendor's namespace (e.g., @runapi.ai/minimax-h3).
Documentation links point to the runapi.ai domain and the vendor's GitHub organization.

minimax-h3