landing-page-design
Warn
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users to install an external package 'belt-sh/cli' using 'npx skills add'. It also references other external skills such as 'inference-sh/skills@ai-image-generation' and 'inference-sh/skills@web-search'. These external dependencies introduce supply chain risks as they are not from a pre-verified trusted source.
- [COMMAND_EXECUTION]: The skill uses the 'belt' CLI to execute various subcommands, including 'belt login' and 'belt app run'. These commands involve authenticating with and executing tasks on remote infrastructure provided by 'inference.sh'.
- [REMOTE_CODE_EXECUTION]: By instructing the agent to run remote applications through the 'belt' tool (e.g., 'falai/flux-dev-lora', 'tavily/search-assistant'), the skill facilitates the execution of code on external servers, processing locally provided inputs.
- [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface for indirect prompt injection. It ingests user-provided descriptions for landing pages and interpolates them directly into prompts sent to remote image generation and search services without explicit boundary markers or sanitization steps.
Audit Metadata