skills/minimax-ai/cli/mmx-cli/Gen Agent Trust Hub

mmx-cli

Pass

Audited by Gen Agent Trust Hub on Apr 25, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Recommends the global installation of the 'mmx-cli' package from npm to enable the skill's functionality.
  • [COMMAND_EXECUTION]: Utilizes the 'mmx' command-line tool to perform content generation, resource management, and search queries.
  • [DATA_EXFILTRATION]: Accesses sensitive local credential files at '/.mmx/credentials.json' and '/.mmx/config.json' for platform authentication.
  • [DATA_EXFILTRATION]: Executes network operations to communicate with the MiniMax API and fetches data from external, user-specified URLs in commands like 'music cover' and 'vision describe'.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection by processing untrusted data from various external sources.
  • Ingestion points: Receives external input through local files via '--messages-file', remote image/audio URLs, and results from web search queries.
  • Boundary markers: The skill documentation does not define specific delimiters or instructions to ignore embedded commands in the untrusted data.
  • Capability inventory: The skill possesses the ability to execute shell commands ('mmx'), perform network requests, and write to the file system using '--out' or '--download' flags.
  • Sanitization: There are no documented mechanisms for sanitizing or validating input from untrusted files or remote URLs.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 25, 2026, 01:26 AM