Skills
Navigate
skills/project-fitness-co/pfc-club-marketing/higgsfield-marketplace-cards/Gen Agent Trust Hub

higgsfield-marketplace-cards

Fail

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructions provide a command to download an installer script from https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh and execute it immediately by piping it to the shell (| sh). This pattern bypasses local verification and executes remote code on the host system.
  • [EXTERNAL_DOWNLOADS]: The skill performs network requests to download external setup scripts from GitHub at runtime.
  • [COMMAND_EXECUTION]: The skill requires access to the Bash tool to perform bootstrapping tasks and to execute the higgsfield CLI for generating image assets.
  • [PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection via the user-supplied prompt argument.
  • Ingestion points: User input is passed to the --prompt parameter of the higgsfield marketplace-cards create command in SKILL.md.
  • Boundary markers: While the prompt is wrapped in shell quotes, there are no logical separators or protective instructions to prevent the model from following malicious directives embedded within the data.
  • Capability inventory: The skill uses the Bash tool to interact with system shell commands and external service APIs.
  • Sanitization: No input validation or sanitization is performed on the user prompt before it is passed to the command-line interface.
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
Risk Level
CRITICAL
Analyzed
May 5, 2026, 10:43 AM
Security Audit — agent-trust-hub — higgsfield-marketplace-cards