guided-social-clip

Warn

Audited by Gen Agent Trust Hub on Feb 22, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill installs npm packages from the @helios-project scope (core, renderer, cli). While required for functionality, these do not belong to the list of trusted organizations and represent a third-party dependency risk.
  • INDIRECT_PROMPT_INJECTION (LOW): In Step 3, the skill fetches content from a user-provided URL to extract brand intelligence. This content is then used to generate creative specifications and code.
  • Ingestion points: External web content fetched via URL in Step 3.
  • Boundary markers: Absent; the agent is not instructed to use delimiters or ignore instructions within the fetched content.
  • Capability inventory: Subprocess execution (npm install, npx helios, ffmpeg) and file system writes (composition.html).
  • Sanitization: Absent; signals from the external site are used directly to drive the generation of the motion design specification and subsequent HTML/JS code.
  • COMMAND_EXECUTION (MEDIUM): The skill implements a dynamic execution pipeline where it generates a local HTML/JavaScript file (composition.html) and executes it via the Helios CLI (npx helios render). Because the generated code is influenced by potentially untrusted data from external URLs, this creates a path for influenced code execution.
  • DATA_EXPOSURE (SAFE): The skill correctly handles sensitive information by instructing the agent not to attempt to read .env files or verify API keys autonomously, relying instead on user input.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 22, 2026, 06:21 PM