higgsfield-youtube-thumbnail
Fail
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill automates the installation of the Higgsfield CLI by downloading and executing a shell script:
curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh. This resource is hosted on the vendor's official GitHub repository. - [COMMAND_EXECUTION]: The agent uses the
higgsfieldCLI to manage authentication, retrieve model configurations, and execute image generation/edit jobs via Bash commands. - [INDIRECT_PROMPT_INJECTION]: The skill accepts untrusted user input for video titles and scene descriptions which are then used to build prompts for image models.
- Ingestion points: User-provided
video-topic-or-title,scene requirements, andheadlinetext inSKILL.md. - Boundary markers: The skill uses a structured
Prompt contractto organize inputs, but does not provide explicit instructions for the image model to ignore instructions potentially embedded in the user text. - Capability inventory: Execution of CLI commands via Bash to generate and modify images.
- Sanitization: No evidence of text sanitization or filtering for the user-provided content before it is piped into the generation commands.
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