landing-page-design
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install an external package 'belt-sh/cli' via npx and references external documentation for installation from a GitHub repository (inference-sh/skills).
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data via the 'tavily/search-assistant' tool.
- Ingestion points: Data from the open web is retrieved via competitor research queries in SKILL.md.
- Boundary markers: The skill lacks explicit delimiters or instructions to the agent to disregard instructions potentially embedded within the search results.
- Capability inventory: The skill has access to the 'belt' command-line tool, which can execute various AI models and applications.
- Sanitization: There is no evidence of sanitization or filtering applied to the external search content before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill uses the 'belt' CLI tool via a Bash tool with broad permissions (belt *). It executes various subcommands to run remote AI applications, passing JSON input which could potentially lead to command injection if user-supplied strings are not properly handled by the underlying CLI.
Audit Metadata