landing-page-design
Pass
Audited by Gen Agent Trust Hub on May 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
beltCLI tool to execute specific applications for generating hero images and performing competitor research (e.g.,falai/flux-dev-lora,tavily/search-assistant). These commands are constrained to the scoped operations allowed for thebelttool.\n- [EXTERNAL_DOWNLOADS]: Includes instructions for the user to install thebeltCLI skill (belt-sh/cli) and provides links to setup documentation on the vendor's GitHub repository (github.com/inference-sh). These downloads are necessary for the skill's primary functionality and originate from the author's infrastructure.\n- [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by retrieving content from external websites via search tools.\n - Ingestion points: Market research data from
tavily/search-assistantandexa/answercalls inSKILL.md.\n - Boundary markers: Absent; there are no specific delimiters or instructions for the agent to treat search results as untrusted content.\n
- Capability inventory: Restricted to the
beltCLI ecosystem; the skill does not have permissions for arbitrary shell execution, sensitive file access, or broad network operations beyond the tool's scope.\n - Sanitization: No explicit sanitization or filtering of external search results is performed before they are processed by the agent.
Audit Metadata