The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill includes a command to install the vendor's CLI tool by piping a remote script to the shell: curl -fsSL https://cli.inference.sh | sh. While this originates from the vendor's own infrastructure, executing remote scripts directly in a shell is a sensitive pattern.
[EXTERNAL_DOWNLOADS]: The skill references binary downloads from dist.inference.sh. The documentation indicates that the installation script performs OS detection and SHA-256 verification of the downloaded files.
[COMMAND_EXECUTION]: The skill utilizes the infsh command via a bash tool for image generation and login actions. It also utilizes npx to add related skills from the inference-sh repository.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection (Category 8).
Ingestion points: Data enters the agent context through the html field in the infsh app run commands.
Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands when the agent populates the HTML templates with user-provided text.
Capability inventory: The skill possesses the capability to execute subprocesses via the infsh CLI and interact with external AI services.
Sanitization: The provided examples do not demonstrate sanitization or escaping of content before it is processed by the HTML-to-image engine.

email-design