The Agent Skills Directory

[SAFE]: No malicious patterns detected. The skill's behavior is consistent with its stated purpose of enhancing UI generation prompts.
[DATA_EXPOSURE]: The skill uses the Read tool to access a project's DESIGN.md file. This is used solely to extract design tokens (colors, typography) to ensure visual consistency in the generated output. No sensitive files or credentials are targeted.
[COMMAND_EXECUTION]: The skill suggests writing the final prompt to a file (e.g., next-prompt.md) using the Write tool. This is a standard workflow for agentic prompt loops and does not involve the execution of arbitrary system commands.
[EXTERNAL_DOWNLOADS]: The skill references official documentation at stitch.withgoogle.com. This is a trusted domain associated with the technology the skill is designed to support.
[PROMPT_INJECTION]: The skill acts as a transformer for user input. While it ingests untrusted data to generate prompts (creating an indirect prompt injection surface), it does so within a highly structured template that limits the risk of instruction override. No explicit injection or bypass patterns were found in the skill logic.

enhance-prompt