Install
$ agentstack add mcp-wjh1547485653-max-minicpm-vision-mcp ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
MiniCPM Vision MCP 🎬👁️🎵
为 DeepSeek V4.0 等单模态大模型 装上眼睛和耳朵 —— 本地运行的视觉+音频 MCP 服务器,基于 Ollama + MiniCPM-V 4.6 + faster-whisper。
> 图片描述 · 视频帧分析 · 音频语音转文字 · 完全本地 · 数据不出机
✨ 为什么选这个?
| | 本项目 | xkiranj/ollama-vision-mcp | MikeyBeez/mcp-vision | |---|---|---|---| | 图片识别 | ✅ | ✅ | ✅ | | 视频分析 | ✅ 唯一!| ❌ | ❌ | | 音频转录 | ✅ 唯一!| ❌ | ❌ | | 中文模型 | ✅ MiniCPM 原生中文 | ❌ llava 英文为主 | ❌ | | 安装 | 1 文件,丢进去就行 | Python venv + pip | macOS 限定 | | Windows | ✅ | ⚠️ | ❌ | | 自动清理 | ✅ 分析完自动删视频 | ❌ | ❌ |
🚀 快速开始
1. 安装 Ollama 并拉取模型
ollama pull minicpm-v4.6
2. 安装依赖
npm install @ffmpeg-installer/ffmpeg @modelcontextprotocol/sdk
pip install faster-whisper
3. 配置 AI Agent
Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"vision": {
"command": "node",
"args": ["/path/to/vision_mcp_server.mjs"]
}
}
}
Reasonix (reasonix.toml):
[[plugins]]
name = "vision"
command = "node"
args = ["/path/to/vision_mcp_server.mjs"]
重启客户端即可使用。
🛠️ 工具
describe_image — 描述图片
输入: 图片文件的绝对路径
输出: 中文图片内容描述
describe_video — 分析视频
输入: 视频文件路径 或 HTTP/HTTPS URL, interval (可选)
输出: 时间线帧描述 + AI 总结 + 自动清理
describe_audio — 音频转文字 🆕
输入: 视频/音频文件路径 或 URL
输出: 音频元数据 + faster-whisper 语音转文字
📊 视频分析流程
URL/文件 → 下载 → FFmpeg 提取关键帧 → MiniCPM-V 逐帧分析 → 汇总总结 → 自动清理
🔧 依赖
- Ollama + MiniCPM-V 4.6
- FFmpeg via
@ffmpeg-installer/ffmpeg - faster-whisper for audio transcription
- @modelcontextprotocol/sdk
🌍 适用场景
- 📸 让 DeepSeek 等纯文本模型看懂截图和照片
- 🎬 快速了解视频内容,不用逐秒观看
- 🎵 提取视频中的语音并转文字
- 📱 分析抖音/B站等平台视频
- 🔒 敏感图片/视频本地处理,数据不出机
- 🇨🇳 中文内容理解(MiniCPM 原生中文)
📄 许可证
MIT License
🙏 致谢
- MiniCPM-V · Ollama · faster-whisper · MCP
Source & license
This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: wjh1547485653-max
- Source: wjh1547485653-max/minicpm-vision-mcp
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.