The Agent Skills Directory

[SAFE]: The skill implements a version check via a local script scripts/update-check.sh. This script fetches a version string from a trusted vendor repository on GitHub, validates the format using a strict regular expression, and does not execute the downloaded content.
[SAFE]: External dependencies and network operations are restricted to the Krea.ai platform's official tools and APIs, ensuring that all operations remain within the vendor's scoped environment.
[SAFE]: Prompting instructions are focused on creative collaboration and model selection, following standard guidelines without attempting to bypass security filters or extract sensitive system information.
[SAFE]: The use of the bash utility is limited to running the internal update check script provided with the skill, which is a benign operational task.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user prompts and images for generation tasks, creating a standard attack surface for indirect injection. However, the instructions focus on structured output and vision-based QA, and the risk is considered low and inherent to the skill's primary purpose.

krea-generate