jimeng-cli-text2image

Fail

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches the dreamina CLI tool directly from the official ByteDance Jimeng domain (jimeng.jianying.com) during the installation phase. \n- [REMOTE_CODE_EXECUTION]: The skill provides instructions to install the utility using a shell script piped from the vendor's official domain via curl | bash, which constitutes a remote code execution pattern for environment setup. \n- [COMMAND_EXECUTION]: Executes several terminal commands associated with the dreamina utility, including text2image, user_credit, login, and query_result, to interact with the Jimeng image generation service. \n- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by interpolating untrusted user data into shell command arguments. \n
  • Ingestion points: User-provided text prompts are passed to the --prompt argument in the text2image command within SKILL.md and related examples. \n
  • Boundary markers: Commands are structured to wrap user input in double quotes (e.g., --prompt="<prompt>"), providing a basic structural delimiter. \n
  • Capability inventory: The skill utilizes terminal execution capabilities to run the vendor's binary and handle task polling. \n
  • Sanitization: No specific shell-escaping or content-validation logic is included in the instructions to sanitize the user-provided prompt before execution.
Recommendations
  • HIGH: Downloads and executes remote code from: https://jimeng.jianying.com/cli - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 7, 2026, 10:13 AM
Security Audit — agent-trust-hub — jimeng-cli-text2image