Atlas / MCP servers / frank-ay / Mimo

MimoSAFE

Name: Mimo
Author: frank-ay

mcp/frank-ay/mimo

MCP server by frank-ay · claude code · codex · fastmcp · Audited 2026-10-08

⚑ Report / re-scan Source ↗

把小米 MiMo 全模态能力(对话/图像/视频/TTS/声音克隆)封装成 stdio MCP Server,Claude Code 与 Codex 可直接调用

Verdict

SAFE

Grade

Trust score

89 /100

Exposed tools

6 3r · 2w · 1d

Transport

—

License

MIT

Stars

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

[](https://www.python.org/) [](LICENSE) [](https://modelcontextprotocol.io/) [](https://github.com/Frank-ay/mimo-mcp/releases/latest)

上图为 60 秒预览(GIF,自动循环);完整 108 秒含音频版: 📥 mimo-mcp-intro.mp4(3.2 MB)

把小米 MiMo 的全模态能力——多模态对话 / 图像 / 视频理解 / TTS / 声音克隆 / 声音设计 / ASR——封装为一个 stdio MCP Server,让 Claude Code 与 Codex 等编程工具直接当 tool 调用,并附带一个本地 Web 管理面板(9 个页面:概览 / 沙盒 / 文字转语音 / 图像视频 / 音色库 / 克隆 / 设计 / 语音转写 / 审计)。

亮点

一处实现,Claude Code & Codex 都能用
11 个 MCP tool 覆盖 F1-F8 全模态能力
本地 SQLite 持久化音色库,克隆 / 设计的 voice 一次创建到处用
mimo.tts 自动按 voice 类型路由到对应 MiMo 模型(default → tts / clone → voiceclone / design → voicedesign)
含 Token Plan 套餐适配 + thinking 模型 max_tokens 兜底等踩坑经验
Web 控制台支持长文按句末标点自动切段批量合成(SSE 流式渲染)
TTS 支持 v2.5 自然语言风格指令(导演模式)与音频标签;ASR(mimo-v2.5-asr)中英双语 + 方言识别、自动加标点

1. 三步上手

1.1 安装依赖

# 后端(Python 3.11+ + uv)
uv sync

# 前端(Node 22+ + pnpm)
cd webui/frontend && pnpm install && cd ../..

1.2 配置 API Key

cp .env.example .env
# 用编辑器打开 .env,把 MIMO_API_KEY 改成你在 platform.xiaomimimo.com 拿到的真实 key

1.3 启动

最快:一条命令同时拉起 Web 前后端(开发模式),Ctrl+C 一起停:

./scripts/dev.sh        # 后端 :7801 + 前端 :5173(Vite HMR)

打开即可看到控制台。

更多启动方式:

Read from source at commit 5713940e77d2OBSERVED · 2026-10-08

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.

claude-code

claude mcp add mimo-mcp-webui --env MIMO_API_KEY=${MIMO_API_KEY} -- npx -y [email protected]

claude-desktop

{
  "mcpServers": {
    "mimo-mcp-webui": {
      "command": "npx",
      "args": [
        "-y",
        "[email protected]"
      ],
      "env": {
        "MIMO_API_KEY": "${MIMO_API_KEY}"
      }
    }
  }
}

Exposed tools (6)

3 read · 2 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

Tool	Risk	Description
`mimo.health`	read	result = await api_usage.health_check()
`mimo.image_understand`	read	inputs = [ImageInput(**i) for i in images]
`mimo.usage`	read	return await api_usage.usage_summary(since_hours, _get_storage())
`mimo.voice_clone_create`	write	req = VoiceCloneCreateRequest(
`mimo.voice_delete`	destructive	ok = await _get_storage().delete_voice(voice_id)
`mimo.voice_design_create`	write	req = VoiceDesignCreateRequest(

Trust audit

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

Layer	What it checks	Result
L0	Provenance & inventory	PASS
L1	Static analysis of the code	PASS
L2	Instruction surface (what it tells the agent)	PASS
L3	Class-specific surface	WARN
L4	Behavioural (sandbox)	SKIPPED

What the source does

Filesystem: declared (2 observation(s))
Network: declared (7 observation(s))
Shell: none-observed
Dependencies: not all pinned
Secrets in source: none-found

Findings (11)

MEDIUMFilesystem / path · mcp.destructive_tools · CWE-22, CWE-59

mimo.voice_delete

Why it matters. 1 tool(s) can delete or overwrite

Fix. prefer a read-only mode or scoped tokens; the page states the blast radius