The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions guide the agent to construct shell commands using user-supplied text for synthesis. This pattern creates a vulnerability surface where malicious input could attempt to escape the command structure.
Ingestion points: User-provided strings for the text field in shell command arguments (SKILL.md).
Boundary markers: No explicit delimiters or sanitization instructions are provided in the examples.
Capability inventory: The skill utilizes the Bash tool to execute belt CLI operations.
Sanitization: The skill lacks instructions for escaping special characters or validating user input before it is passed to the tool.
[EXTERNAL_DOWNLOADS]: The skill references external resources and installation scripts required for its operation.
Evidence: Links to installation instructions at https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md and references the belt-sh/cli skill via npx.
Context: These references point to established services and are consistent with the stated purpose of integrating with the inference.sh platform.

text-to-speech