higgsfield-websites

Fail

Audited by Gen Agent Trust Hub on Aug 8, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions to download and execute a CLI installation script from the author's official GitHub repository.
  • Evidence: curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh in SKILL.md.
  • This script is used to set up the necessary environment for the website builder.
  • [REMOTE_CODE_EXECUTION]: A bundled Python utility script is designed to download fonts and brand assets from a vendor-controlled server to the local file system.
  • Evidence: The ensure_asset function in references/app-cover.md fetches files from https://static.higgsfield.ai/website-builder/app-cover-generator.
  • The script uses urllib.request to retrieve these resources and open for local caching.
  • [COMMAND_EXECUTION]: The skill includes shell scripts for local processing of media assets, such as video encoding and frame extraction.
  • Evidence: scroll-scrub-video.sh in references/scroll-scrub-asset-video.md executes ffmpeg with various flags to prepare video content.
  • [PROMPT_INJECTION]: The skill's workflow involves ingesting user-supplied descriptions to generate prompts for various AI models, which creates a surface for indirect prompt injection.
  • Ingestion points: User-provided website concepts and brand briefs in SKILL.md and references/website-flow.md.
  • Boundary markers: Not explicitly implemented in the provided code templates or instructions.
  • Capability inventory: The skill has access to shell execution via the Bash tool and file system writes for project scaffolding.
  • Sanitization: No input filtering or prompt escaping logic is present in the analyzed files.
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
Aug 8, 2026, 04:42 PM
Security Audit — agent-trust-hub — higgsfield-websites