The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the run_command tool to execute shell commands for checking the environment and installing dependencies.
[REMOTE_CODE_EXECUTION]: The skill contains a detailed Python script within SKILL.md that it instructs the agent to execute dynamically at runtime using run_command. This script handles image processing and text rendering based on user-supplied variables.
[EXTERNAL_DOWNLOADS]: The skill attempts to download and install the Pillow package from the Python Package Index (PyPI), which is a well-known and official package registry.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. Ingestion points: User-provided product images and text descriptions from the conversation. Boundary markers: Absent; the user text is directly interpolated into prompts and scripts without delimiters. Capability inventory: The skill uses run_command (Python execution), generate_image, and browser_subagent tools. Sanitization: No validation or escaping is applied to user-controlled text before it is used to generate image prompts or rendered via the Python PIL script.

goods-images