Skills
Navigate
skills/opencue/skills/higgsfield-to-medusa-products/Gen Agent Trust Hub

higgsfield-to-medusa-products

Fail

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill's bootstrap instructions in SKILL.md direct the user to download and execute a script directly from a remote source using a dangerous pattern: curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh. Executing remote scripts by piping them directly to a shell can lead to full system compromise if the repository or communication is compromised.
  • [EXTERNAL_DOWNLOADS]: The skill is designed to download binary tools (Higgsfield CLI) and fetch image assets from the Higgsfield CDN at runtime.
  • [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface where product prompts defined in an external manifest.json are passed to the Higgsfield CLI without sanitization. Maliciously crafted prompts in the manifest could manipulate the generation process.
  • Ingestion points: The shots[].prompt field in the manifest JSON.
  • Boundary markers: None identified.
  • Capability inventory: Uses higgsfield CLI, aws CLI, and curl for API interactions in scripts/run-pipeline.sh.
  • Sanitization: Values are parsed via jq but injected directly into CLI arguments.
  • [COMMAND_EXECUTION]: The skill relies on complex bash scripts (scripts/run-pipeline.sh and scripts/load-env.sh) to orchestrate its workflow, which involves executing multiple external CLI utilities and managing local environment files.
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
HIGH
Analyzed
Jun 19, 2026, 11:49 PM
Security Audit — agent-trust-hub — higgsfield-to-medusa-products