higgsfield-video-explainer

Fail

Audited by Gen Agent Trust Hub on Sep 26, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions to download and execute a setup script directly from the vendor's official code repository to install the Higgsfield CLI tool.
  • Evidence: The command curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh is used in the Bootstrap section of SKILL.md.
  • [EXTERNAL_DOWNLOADS]: Fetches the Higgsfield CLI installer and allows for downloading external web images to be used as style references for video generation.
  • [COMMAND_EXECUTION]: Extensively uses the Bash tool to interact with the Higgsfield CLI for model resolution, voice listing, media generation, and server-side assembly.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to create narration scripts and visual prompts, creating a vulnerability surface.
  • Ingestion points: Processes user-provided Topic or personal/philosophical story text and Optional local source documents in SKILL.md, along with data retrieved during the `Phase R
  • research` step.
  • Boundary markers: There are no explicit delimiters or system-level instructions to ignore embedded commands within the processed source material.
  • Capability inventory: The skill can execute shell commands via Bash, perform network operations via curl and the Higgsfield CLI, and generate files.
  • Sanitization: No validation or sanitization of input documents or research data is performed before they are used to generate instructions for the media models.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 26, 2026, 06:53 PM