higgsfield-product-photoshoot

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides a bootstrap instruction to download and execute an installation script from the author's official GitHub repository (https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh). This is the standard installation procedure for the Higgsfield CLI tool.
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool to run commands for authentication (higgsfield auth login) and image generation (higgsfield product-photoshoot create). These commands are integral to the skill's stated purpose of creating brand-quality visuals.
  • [INDIRECT_PROMPT_INJECTION]: The skill accepts user-supplied text prompts and image references which are passed to the CLI for processing.
  • Ingestion points: User-provided descriptions and file paths are ingested in SKILL.md through a structured interview process.
  • Boundary markers: The instructions explicitly state that the backend assembles the final prompt, preventing the agent from freehanding instructions to the image model.
  • Capability inventory: The skill is permitted to use the Bash tool to interact with the Higgsfield CLI.
  • Sanitization: The skill relies on the vendor's backend prompt enhancer to structure the final input, reducing the attack surface for prompt-based manipulation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 06:04 PM
Security Audit — agent-trust-hub — higgsfield-product-photoshoot