The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains explicit instructions for the agent to execute system commands with administrative privileges (sudo apt update && sudo apt install python3). This facilitates potential privilege escalation and unauthorized system modifications.
[COMMAND_EXECUTION]: The core functionality of the skill involves executing a local Python script (search.py) with parameters derived from user input. This pattern presents a risk of command injection if the underlying script does not adequately sanitize the provided arguments.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to download and install external software packages from system repositories using tools like brew, apt, and winget.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting untrusted user input via search queries, processing it, and then instructing the agent to persist and later retrieve this data as an authoritative 'Source of Truth' for generating UI code.
Ingestion points: User-supplied search queries passed to the search.py script.
Boundary markers: Absent from the provided instruction workflow.
Capability inventory: Shell command execution (sudo, brew, winget, python) and file system writes (--persist flag).
Sanitization: No sanitization or validation of the input query is mentioned in the workflow.

ui-ux-pro-max