什么是 MiniMax to T@elegrimm 媒体套件？

MiniMax to T@elegrimm 技能在尖端生成式 AI 与即时通讯工作流之间架起了一座无缝桥梁。通过将 MiniMax 的多模态能力集成到您的 Openclaw Skills 设置中，您可以以编程方式创建图像、语音和视频。该工具专门用于处理经过身份验证的媒体 URL 复杂性，确保生成的每个资产都能在 T@elegrimm 生态系统中立即共享和访问，使其成为您 Openclaw Skills 库中至关重要的一部分。

下载入口:https://github.com/openclaw/skills/tree/main/skills/hoyin258/minimax-to-telegram

安装与下载

1. ClawHub CLI

从源直接安装技能的最快方式。

2. 手动安装

将技能文件夹复制到以下位置之一

优先级：工作区 > 本地 > 内置

3. 提示词安装

将此提示词复制到 OpenClaw 即可自动安装。

MiniMax to T@elegrimm 媒体套件 应用场景

• 使用 Openclaw Skills 自动创建 AI 生成的每日简报并发送至 T@elegrimm 群组。
• 在您的工作流中使用 MiniMax 生成工具快速制作营销视觉原型。
• 针对本地化粤语受众的自定义克隆语音消息互动。
• 编排复杂的视频生成流水线，并在完成后通过聊天通知相关方。

1. 用户或代理在 Openclaw Skills 框架内触发特定的媒体生成工具。
2. 请求通过 mcporter 路由至 MiniMax MCP 服务器，并带有所需的提示词和参数。
3. MiniMax 生成资产并返回包含必要身份验证令牌的完整签名 URL。
4. 技能逻辑捕获该 URL 并准备 T@elegrimm 消息载荷，以确保在 Openclaw Skills 内完成交付。
5. 集成的消息工具将媒体发送至目标 ch@t_id，包括任何所需的说明文字或音频文本。

MiniMax to T@elegrimm 媒体套件 配置指南

要开始在 Openclaw Skills 中使用此集成，首先安装 mcporter CLI：

npm install -g mcporter

接下来，配置您的 API 凭据，并在位于 ~/.mcporter/config.json 的配置文件中将资源模式设置为 URL：

{
  "env": {
    "MINIMAX_API_KEY": "您的 API 密钥",
    "MINIMAX_RESOURCE_MODE": "url"
  }
}

最后，将 MiniMax 服务器添加到您的本地环境以启用这些 Openclaw Skills：

mcporter mcp add minimax-mcp

MiniMax to T@elegrimm 媒体套件 数据架构与分类体系

该技能作为 Openclaw Skills 生态系统的一部分，管理多个生成类别的数据：

name: minimax-to-telegram
description: |
  Generate images, audio, video using MiniMax MCP and send to T@elegrimm. 
  Use when user wants to create media with MiniMax and deliver it via T@elegrimm.

Setup (Prerequisites)

1. Install mcporter

# 如果未有 npm/npx
npm install -g mcporter

或者用 npx 直接跑：

npx mcporter --help

2. Set MiniMax API Key

响 terminal 度 set 環境變數：

export MINIMAX_API_KEY="your-api-key-here"

或者响 ~/.mcporter/config.json 入面 set：

{
  "env": {
    "MINIMAX_API_KEY": "your-api-key-here",
    "MINIMAX_RESOURCE_MODE": "url"
  }
}

3. Add MiniMax MCP Server

mcporter mcp add minimax-mcp

MiniMax MCP Skill

Use mcporter to call MiniMax MCP server tools.

Prerequisites

• mcporter CLI installed
• MiniMax MCP server configured in mcporter

Available Tools

Basic Usage

Image Generation

mcporter call minimax-mcp.text_to_image prompt:"your prompt" aspectRatio:"4:3"

Audio Generation (TTS)

mcporter call minimax-mcp.text_to_audio text:"Hello world" voiceId:"male-qn-qingse"

Video Generation

mcporter call minimax-mcp.generate_video prompt:"your video description"

Sending to T@elegrimm

IMPORTANT: When MiniMax returns a URL, it includes a query string with authentication token. You MUST use the FULL URL including all query parameters.

Correct (full URL with token):

?Expires=xxx&OSSAccessKeyId=xxx&Signature=xxx

Incorrect (URL without token):

Sending Image to T@elegrimm

1. Call text_to_image and capture the FULL URL (including query string)
2. Send directly to T@elegrimm using message tool with media parameter:

message(
  action="send",
  channel="telegram",
  target="",
  media="",
  message="Your caption"
)

Sending Audio to T@elegrimm

Same approach - use FULL URL with token:

message(
  action="send",
  channel="telegram",
  target="",
  media="",
  message="Your caption"
)

IMPORTANT: When sending audio, ALWAYS include the text content as the message caption below the audio!

Example:

# Generate audio
audio_url = "?Expires=xxx&Signature=xxx"
text_content = "呢段係你想既文字內容..."

# Send with both audio and text
message(
  action="send",
  channel="telegram",
  target="",
  media=audio_url,
  message=text_content  # Always include the text!
)

Common Parameters

text_to_image

• prompt: Text description of desired image
• aspectRatio: "1:1", "16:9", "4:3", "3:2", "2:3", "3:4", "9:16", "21:9"
• n: Number of images (1-9)
• model: Model to use (default: "image-01")

text_to_audio

• text: Text to convert to speech
• voiceId: Voice ID (e.g., "male-qn-qingse", "female-shaonv")
• speed: Speech speed (0.5-2)
• emotion: "happy", "sad", "angry", "fearful", "disgusted", "surprised", "neutral"
• format: "mp3", "wav", "pcm", "flac"
• languageBoost: Enhance recognition for specific languages/dialects
  • Cantonese: "Chinese,Yue" or "Chinese"
  • Mandarin: "Chinese" or "Chinese,Mandarin"
  • English: "English"
  • Always include this when generating Cantonese audio!

generate_video

• prompt: Video description
• model: "T2V-01", "T2V-01-Director", "I2V-01", "I2V-01-Director", "I2V-01-live", "MiniMax-Hailuo-02"
• duration: 6 or 10 (for MiniMax-Hailuo-02)
• resolution: "768P", "1080P"

Optimization & Troubleshooting

• Timeout Management: Video generation can take significant time (up to 20-30 minutes for high-quality 10s clips). Always pass a large --timeout value (e.g., --timeout 1800000) to mcporter to prevent early termination.
• Gateway Turn Limit: Most OpenClaw profiles have a 10-minute turn timeout (600000ms). To avoid being killed by the gateway, ALWAYS run generate_video with background: true or inside a background process.
• Model Choice: Use MiniMax-Hailuo-02 for higher quality 10-second videos.

Error Handling

If you get 403 Forbidden when sending to T@elegrimm:

• Make sure you're using the FULL URL with query string
• The token (Signature) expires - regenerate if needed

Notes

• MiniMax MCP must be configured with MINIMAX_RESOURCE_MODE=url in mcporter config
• Generated media URLs include authentication tokens that expire
• Always use the complete URL returned by MCP calls

Cantonese (廣東話) Tips

When generating Cantonese audio:

• Use voice ID from Cantonese category (e.g., "Cantonese_PlayfulMan", "Cantonese_CuteGirl")
• Always add: languageBoost:"Chinese,Yue" or languageBoost:"Chinese"
• Example:

  mcporter call minimax-mcp.text_to_audio text:"新年快樂" voiceId:"Cantonese_PlayfulMan" languageBoost:"Chinese,Yue"