ui-ux-pro-max
Warn
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: MEDIUMPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PRIVILEGE_ESCALATION]: The installation instructions in
SKILL.mdsuggest the use ofsudo apt update && sudo apt install python3. Providing the agent with instructions that involve superuser privileges increases the risk of system-wide changes that may not be fully intended or reviewed by the user. - [INDIRECT_PROMPT_INJECTION]: The skill processes user queries and pulls data from local CSV files in the
data/directory. This pattern introduces a vulnerability to indirect prompt injection if the database content is compromised or contains instructions designed to manipulate the agent. - Ingestion points: User input is passed to
scripts/search.py, which subsequently reads content from multiple local CSV files. - Boundary markers: The script provides structured output using Markdown, but it lacks specific instructions or markers to prevent the agent from executing instructions potentially hidden in the search results.
- Capability inventory: The skill uses Python to read local data files; it does not contain network request functionality or file-writing logic.
- Sanitization: The
core.pyscript filters the search query for alphanumeric characters, but it does not sanitize or escape the data retrieved from the CSV files before passing it to the agent. - [COMMAND_EXECUTION]: The skill relies on the execution of a local Python script (
scripts/search.py) via the shell to perform its core functionality. - [EXTERNAL_DOWNLOADS]: The skill references commands to install prerequisites via system package managers (Homebrew, WinGet, and APT), which fetch software from established, official repositories.
Audit Metadata