MiniMax

Build with MiniMax text, speech, video, and music APIs using model routing, compatible SDKs, and safer multimodal workflows.

This skill appears coherent with its stated MiniMax integration purpose, but before installing: 1) Only provide a MINIMAX_API_KEY with appropriately limited scope (use a dedicated key or workspace account you can rotate). 2) Expect the skill to create and read small files under ~/minimax/ — confirm you’re comfortable with that path and that it won't be used for storing large private assets unless you explicitly approve. 3) The skill promises to require explicit approval for private media uploads, voice cloning, enabling remote MCP hosts, or launching paid/long-running jobs — verify the agent actually prompts you for consent during those flows. 4) Review the MiniMax provider docs and any organizational policy about external API keys before using in production. If you want higher assurance, ask the skill author for a minimal example of the exact API calls it will make (base URLs and request shape) or run it with a low-permission test key first.

Type: OpenClaw Skill Name: minimax Version: 1.0.0 The MiniMax skill bundle is a well-structured set of instructions and templates designed to help an AI agent interact with MiniMax's multimodal APIs (text, speech, video, and music). It includes clear safety boundaries, requiring explicit user consent for sensitive actions such as uploading private media, cloning voices, or enabling remote MCP servers. The skill manages its state locally in `~/minimax/` and restricts external communication to official MiniMax endpoints (api.minimax.io) and user-approved hosts, showing no signs of malicious intent or data exfiltration.