higgsfield-youtube-thumbnail
Fail
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill performs remote script execution by downloading an installation script from 'https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh' and piping it directly to the shell ('sh'). This practice executes remote code without local verification of the content.
- [COMMAND_EXECUTION]: The skill uses the Bash tool to run 'higgsfield' CLI commands for account management, model retrieval, and image generation, often using temporary files to pipe complex inputs.
- [INDIRECT_PROMPT_INJECTION]: The skill aggregates user-supplied video details to build prompts for an external generative AI, which is a known attack surface for prompt injection.
- Ingestion points: User-supplied video titles and scene descriptions in SKILL.md.
- Boundary markers: The skill employs prompt structures like 'CHARACTER N:' and 'IMAGE REFERENCES:' to separate inputs.
- Capability inventory: Usage of the Bash tool and Higgsfield CLI for system interaction and remote API calls.
- Sanitization: While it handles shell-safe delivery of prompts through file-piping, it lacks mechanisms to sanitize or filter malicious instructions within the user-provided text content.
Recommendations
- HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata