The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The SKILL.md file contains instructions for the agent to execute sudo apt update && sudo apt install python3 to set up the environment. This pattern requests administrative privileges, which is a high-risk operation for an AI agent.
[COMMAND_EXECUTION]: The workflow in SKILL.md (Steps 2, 2b, 3, and 4) instructs the agent to execute a local Python script search.py using user-provided keywords and product types directly in the command line (e.g., python3 ... "<query>"). This creates a direct command injection vulnerability if the user input contains shell metacharacters like semicolons or backticks.
[PERSISTENCE]: The skill implements a 'Master + Overrides' pattern where it creates and updates files in design-system/MASTER.md and design-system/pages/. These files are intended to be read in future sessions, allowing potentially poisoned or malicious design rules to persist and influence agent behavior across different interactions.
[INDIRECT_PROMPT_INJECTION]: The skill has a defined attack surface for indirect prompt injection.
Ingestion points: User-provided product types, industry names, and keywords are ingested via the search command in SKILL.md.
Boundary markers: None are present; user input is directly interpolated into command strings.
Capability inventory: The skill performs subprocess execution (python3 search.py) and file system writes (design-system/MASTER.md).
Sanitization: There is no evidence of input validation or escaping for the user-provided strings before they are passed to the shell or written to persistent files.

ui-ux-pro-max