meigen-ai-design-mcp
Warn
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSMETADATA_POISONINGINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses
npx -y meigen@1.3.1to download and execute the MCP server code from the npm registry, which is an external and untrusted source. - [METADATA_POISONING]: The skill claims to provide access to non-existent model versions such as 'Midjourney V8.1' (the current official version is 6) and 'GPT Image 2', which may mislead users regarding its true capabilities and the safety of the underlying models.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided prompts and external data from a curated gallery search, creating an attack surface where malicious instructions could be embedded in design inspiration or keywords.
- [COMMAND_EXECUTION]: The installation and CLI modes rely on executing shell commands via
npxto interact with the generation engine. - [DATA_EXFILTRATION]: The skill is designed to read local image files (e.g., via
referenceImagePath) and upload them to the 'MeiGen Cloud' API for processing.
Audit Metadata