The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the belt CLI tool to execute image generation tasks. This usage is appropriately restricted in the metadata via allowed-tools: Bash(belt *), ensuring the agent cannot execute arbitrary shell commands.
[EXTERNAL_DOWNLOADS]: The skill references official installation guidelines and documentation from the inference.sh GitHub repository and website. These references are standard for service integration and do not involve unverifiable remote code execution.
[INDIRECT_PROMPT_INJECTION]: The skill takes user-provided text prompts to generate images, which represents a standard vulnerability surface for generative AI tools.
Ingestion points: User prompts are ingested as arguments for the belt app run command in SKILL.md.
Boundary markers: User input is structured within JSON objects passed as command-line arguments.
Capability inventory: The skill is limited to executing the belt command as defined in the allowed-tools section of SKILL.md.
Sanitization: Input validation and safety filtering are primarily managed by the external inference.sh API and the underlying AI models.

ai-image-generation