minimax-multimodal-toolkit
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFEDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
mmx-clitool using the commandnpm install -g mmx-clito provide its core functionality. - [COMMAND_EXECUTION]: The skill relies on executing the
mmxcommand-line utility for all AI operations, including generation and transcription tasks. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as images, audio files, and text documents which are passed to LLMs and VLMs.
- Ingestion points: Untrusted data is ingested via flags such as
--messages-file,--image,--file, and--text-fileinSKILL.md. - Boundary markers: No specific boundary markers or "ignore" instructions are used in the provided examples to prevent the model from following instructions embedded in the input data.
- Capability inventory: The skill can read local files, write files to the system (e.g., via
--out,--out-dir, or--download), and perform network API calls. - Sanitization: There is no evidence of sanitization or content filtering of the input files before they are processed by the tool.
- [CREDENTIALS_UNSAFE]: The skill manages MiniMax API keys and OAuth tokens, persisting them to the local filesystem at
~/.mmx/config.jsonand~/.mmx/credentials.jsonrespectively. - [DATA_EXFILTRATION]: The skill supports sending data to external webhooks through the
--callback-urlflag in thevideo generatecommand and can fetch content from arbitrary URLs provided in the--imageflag.
Audit Metadata