minimax-understand-image
Pass
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches the 'uv' installation script from 'astral.sh', which is a well-known and trusted source for Python development tools.
- [REMOTE_CODE_EXECUTION]: It uses 'uvx' to dynamically download and execute the 'minimax-coding-plan-mcp' package, which is necessary for the skill's primary function of image understanding.
- [COMMAND_EXECUTION]: Shell commands are used to check for tool availability, create configuration directories at '~/.openclaw/config', and manage the storage of the MiniMax API key.
- [PROMPT_INJECTION]: The skill contains an indirect injection surface as it processes untrusted input from 'sys.argv' (Ingestion points) in 'scripts/understand_image.py'. No boundary markers are present to delimit user input (Boundary markers: absent). The skill has the capability to spawn subprocesses and access the network via the MCP server (Capability inventory). No input sanitization or filtering is performed on the user-provided prompt or image source (Sanitization: absent).
Audit Metadata