M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
M

MiniMax MCP

Official
Official MiniMax MCP server for text-to-speech, image generation, and video generation.
How to install ↓ Compare GitHub source repository ↗
Category
Dev Tools #158 of 438
Stars
★ 1.6k Very popular
Transport
stdio (local process) · SSE
Runtime
Python
Credentials
API key / credential required
License
MIT
Last commit
Tools
8
57FMRS · C

MiniMax MCP server is officially maintained, offers multimodal generation capabilities, easy setup, but requires attention to API key region alignment and potential costs.

Strongest · Documentation 14/20 Weakest · Reliability 8/20

Reliability
8/20
Security and permissions
10/20
Maintenance
13/20
Documentation
14/20
Setup experience
12/20
Why each score
Reliability 8/20
The README describes 8 tools (text_to_audio, list_voices, voice_clone, generate_video, text_to_image, query_video_generation, music_generation, voice_design) with a consistent install/config story, so the happy path is plausible. However, only the README and repository metadata were supplied; the server manifest is not cached and no source files, CI workflows, or committed tests are available to verify that declared tools match real behavior or that the init/tool-list handshake completes. Per static calibration, missing execution evidence lowers the score; error handling and edge cases are also unverifiable.
Security and permissions 10/20
Credentials are handled via environment variables (MINIMAX_API_KEY), the example config uses a placeholder rather than a real token, and the README warns that using these tools may incur costs and that the API key must match the regional host — all positive findings. No red-line issues (malware, credential theft, covert exfiltration, real secrets in docs) appear in the supplied evidence. Deductions: there is no source-level evidence of least-privilege design, no explicit confirmation step for paid/generative operations, and no disclosure of what data is transmitted to MiniMax cloud APIs beyond the cost warning.
Maintenance 13/20
The repository is not archived, has an MIT license, and belongs to the official MiniMax-AI organization; release notes dated July 2, 2025 (voice_design, Hailuo-02, music-1.5) indicate recent active development with a clear ownership and update path. Deductions: no supplied evidence of commit cadence, issue-response quality, dependency update policy, or a security-response channel, and the 41 open issues cannot be assessed without issue-tracker data, so full marks for sustained governance are not justified.
Documentation 14/20
The README covers installation (uv/uvx), authentication (API key acquisition and global/mainland host mapping), client configuration examples for Claude Desktop and Cursor, a tool description table, release notes, screenshot-based usage examples, cost warnings, and a FAQ addressing 'invalid api key' and 'spawn uvx ENOENT'. Deductions: no detailed per-tool parameter schemas, no rate-limit or error-code documentation, limited disclosure of model/tool limitations, and no layered docs beyond the README, so hidden assumptions remain for advanced usage.
Setup experience 12/20
The path to a working connection is concise: obtain an API key, install uv/uvx, and add the provided JSON snippet to a mainstream client; the README even offers a fix for the common 'spawn uvx ENOENT' failure. Per static calibration, setup cannot exceed 15 without verifiable CI/tests in the supplied files; here no manifest, CI, or test execution evidence was provided, so the score is capped well below 15, and the installability of the minimax-mcp package via uvx is unverified in this static review.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessWrites / deletes local filesUses the network

Best for

  • Developers and content creators who need to integrate AI voice, image, and video generation quickly
  • Users who want to use MiniMax APIs directly from MCP-compatible clients

Not for

  • Users who do not need AI generation features
  • Users without a valid API key
  • Users who want offline operation or different models

Required permissions

  • Requires a valid MiniMax API key
  • May incur API usage costs
  • Can read local files or URL resources as input

Risks and side effects

  • API key mismatch or region mismatch can cause 'invalid api key' errors
  • Usage may incur costs
  • Generated content may be subject to copyright and ethical limitations

Setup

Before you start

Runtime：Python

MINIMAX_API_KEY requiredsecret API key from the MiniMax platform user center (Global or Mainland); must match the API host region.
MINIMAX_MCP_BASE_PATH required Local output directory for generated audio/image/video files, e.g. /User/xxx/Desktop.
MINIMAX_API_HOST required API host: https://api.minimax.io (Global) or https://api.minimaxi.com (Mainland); must match the key's region.
Other optional settings (1)
MINIMAX_API_RESOURCE_MODE optional Optional, [url|local], url is default; whether audio/image/video are returned as URLs or downloaded locally.
  1. Obtain an API key from MiniMax (note regional differences).
  2. Install uv (Python package manager).
  3. Configure your MCP client (e.g., Claude Desktop) by adding the server config to claude_desktop_config.json.
claude_desktop_config.json 
{
  "mcpServers": {
    "MiniMax": {
      "command": "uvx",
      "args": [
        "minimax-mcp",
        "-y"
      ],
      "env": {
        "MINIMAX_API_KEY": "insert-your-api-key-here",
        "MINIMAX_MCP_BASE_PATH": "local-output-dir-path, such as /User/xxx/Desktop",
        "MINIMAX_API_HOST": "api host, https://api.minimax.io | https://api.minimaxi.com",
        "MINIMAX_API_RESOURCE_MODE": "optional, [url|local], url is default, audio/image/video are downloaded locally or provided in URL format"
      }
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "MiniMax": {
      "command": "uvx",
      "args": [
        "minimax-mcp",
        "-y"
      ],
      "env": {
        "MINIMAX_API_KEY": "insert-your-api-key-here",
        "MINIMAX_MCP_BASE_PATH": "local-output-dir-path, such as /User/xxx/Desktop",
        "MINIMAX_API_HOST": "api host, https://api.minimax.io | https://api.minimaxi.com",
        "MINIMAX_API_RESOURCE_MODE": "optional, [url|local], url is default, audio/image/video are downloaded locally or provided in URL format"
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add MiniMax -e MINIMAX_API_KEY=insert-your-api-key-here -e 'MINIMAX_MCP_BASE_PATH=local-output-dir-path, such as /User/xxx/Desktop' -e 'MINIMAX_API_HOST=api host, https://api.minimax.io | https://api.minimaxi.com' -e 'MINIMAX_API_RESOURCE_MODE=optional, [url|local], url is default, audio/image/video are downloaded locally or provided in URL format' -- uvx minimax-mcp -y

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Check that the 8 tools (text_to_audio, generate_video, etc.) appear in the client's tool list, then call list_voices — if it returns the available voices, the connection works. Note the API key must match the API host region or an 'invalid api key' error occurs.

Troubleshooting

  1. Check that your API key and host are regionally aligned (Global: api.minimax.io, Mainland: api.minimaxi.com)
  2. If 'spawn uvx ENOENT' occurs, confirm the absolute path of uvx and update the config
  3. Define completion rules when using async video generation

Things to try

Once connected, you can ask your AI assistant things like:

  • Convert this news segment text to speech
  • List all available voices
  • Clone a voice from my audio recording
  • Generate a video from this prompt

Tools 8

text_to_audio writes
Convert text to audio with a given voice.
list_voices read-only
List all voices available.
voice_clone writes
Clone a voice using provided audio files.
generate_video writes
Generate a video from a prompt.
text_to_image writes
Generate an image from a prompt.
query_video_generation read-only
Query the result of a video generation task.
music_generation writes
Generate a music track from a prompt and lyrics.
voice_design writes
Generate a voice from a prompt using preview text.

Use cases

Broadcast a news segment
Clone a voice
Generate a video
Generate images

Supported clients

Claude Desktop
Cursor
Windsurf
OpenAI Agents

Listed from the project's documentation, not tested by this site.

Overview

The MiniMax MCP server is the official Model Context Protocol (MCP) implementation by MiniMax, allowing MCP clients like Claude Desktop, Cursor, Windsurf, and OpenAI Agents to interact with MiniMax's AI generation APIs. It supports text-to-speech, voice cloning, text-to-image, text-to-video (with async query), music generation, and voice design.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 62.9k · Tools 2 Compare with this →
Compare with other MCP servers

Source revision 0856b9aef8a9 Data synced 2026-10-11 Read the FMRS scoring method