The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands using bash to interact with a core dependency script (ofox-image.sh) for image processing and cost estimation.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a processing layer for untrusted user prompts and image files, which are interpolated into commands and API requests.
Ingestion points: User instructions for image modifications and local file paths for image inputs (found in SKILL.md).
Boundary markers: The instructions recommend a specific two-sentence structure for the prompt but do not provide explicit delimiters or "ignore instructions" guardrails for the processed data.
Capability inventory: Shell execution via bash, network communication for API calls (delegated to the core script), and writing edited image files to the local file system.
Sanitization: No explicit sanitization or validation logic is defined within the skill instructions for the user-provided prompt text or image content.
[DYNAMIC_EXECUTION]: The skill implements a bash loop to dynamically resolve the location of the ofox-image.sh script by searching across multiple potential installation paths at runtime.

image-edit