guided-product-demo
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill researches brand information from user-provided URLs. This external content is untrusted and could contain malicious instructions designed to influence the agent's downstream tasks, such as the generation of the creative specification or the Helios composition.
- Ingestion points: Researching user-provided marketing site URLs in Step 3.
- Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the fetched URL content.
- Capability inventory: The skill executes shell commands for package installation, uses
ffmpegfor audio analysis, and runsnpx helios renderto execute generated code. - Sanitization: No sanitization or validation of the fetched URL content is performed before processing.
- [DYNAMIC_EXECUTION]: In Step 8, the skill generates a
composition.htmlfile containing JavaScript and CSS animations. This generated file is subsequently executed via the Helios CLI in Step 9 to render the final video output. - [EXTERNAL_DOWNLOADS]: The skill installs third-party Node.js packages (@helios-project/core, @helios-project/renderer, @helios-project/cli) and interacts with the ElevenLabs API for music generation.
- [COMMAND_EXECUTION]: The skill makes heavy use of system commands including
npm install,ffmpeg, andnpxto facilitate the video production pipeline.
Audit Metadata