What problem does it solve? Testing prompts or comparing models normally requires switching between provider SDKs and writing boilerplate code for each API. This Skill lets you send a raw prompt to any model configured in your MiniMax Code config.yaml with a single command, without touching provider-specific request formats. ## Core Features & Use Cases - Multi-Protocol Support: Automatically maps providers to the correct API protocol — Anthropic Messages, OpenAI Chat Completions, or Gemini generateContent — based on the provider's npm field in config.yaml. - Model Discovery and Selection: List all configured models with --list, call a specific provider/model reference, or fall back to the configured defaultModel. - Streaming and Parameter Control: Supports streaming output, system prompts, temperature, max tokens, custom timeouts, and JSON output for scripting. - Use Case: Compare how gemini/gemini-2.5-pro and an OpenAI-compatible model answer the same prompt by running the script twice with different --model flags, then review the outputs side by side. ## Quick Start Ask the agent to call a configured model with your prompt, for example: run the llm-call script with model gemini/gemini-2.5-pro and the prompt "Summarize this document".
