Multimodal Studio

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided creative briefs to generate prompts for visual models, creating an attack surface for indirect prompt injection.
  • Ingestion points: The 'BRIEF' stage described in SKILL.md.
  • Boundary markers: None identified in the prompt templates.
  • Capability inventory: Tools for generate_image, generate_video, and create_character.
  • Sanitization: No explicit validation or filtering of user-supplied brief content is mentioned before interpolation into generation prompts.
  • [EXTERNAL_DOWNLOADS]: The skill documentation describes connecting to Higgsfield AI's official MCP server (https://mcp.higgsfield.ai/mcp) to provide the underlying generation tools.
  • [COMMAND_EXECUTION]: The model matrix resource provides instructions for local environment configuration and execution of the Higgsfield MCP server using the python -m higgsfield_mcp.server command.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 09:15 AM