ultimate-ui
Warn
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: Automated environment probing. The SKILL.md file instructs the agent to execute a shell command (
ls) targeting specific hidden directories in the user's home folder (~/.claude/skills/,~/.agent/skills/, and~/.agents/skills/) to list installed skills. This metadata gathering occurs silently (using2>/dev/null) upon skill activation without a specific user request. - [PROMPT_INJECTION]: Behavioral override instruction. The skill mandates that the agent must prefix its first response with the 🧢 emoji, which is an attempt to override the standard output format of the AI agent.
- [EXTERNAL_DOWNLOADS]: Remote skill installation. The skill provides instructions and prompts the agent to install additional remote components from the 'AbsolutelySkilled' repository using the
npxcommand, which involves downloading and executing third-party code.
Audit Metadata