SA
SkillAvatars
Back to MCP Tools

luma-mcp

data-ai MCP Server

多模型视觉理解 MCP 服务器，为不支持图片理解的 AI 编码模型提供视觉能力：分析截图、报错、UI 与文档，可接入多家主流视觉大模型。Multi-model vision MCP server that adds image understanding to AI coding models without native vision — analyze screenshots, errors, UI and documents via major vision LLM providers.

byJochenYang
VerifiedFreshInstall ReadyReviewed
data-aidata-ai
19 views114 stars12 forksv1.7.1MIT
View on GitHubHomepage

Why This Matters

Discovered via github-seeds:mcp-hot and last synced 3d ago.

VerifiedFreshInstall ReadyReviewed
Source
github-seeds:mcp-hot
Stars
114
Last synced
3d ago
Install
Instructions detected

Install

1. Install the package

npx -y luma-mcp

2. Add to claude_desktop_config.json

{
  "mcpServers": {
    "luma-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "luma-mcp"
      ]
    }
  }
}

Config file location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) / %APPDATA%\Claude\claude_desktop_config.json (Windows)

33
Tools
0
Resources
0
Prompts
Standard I/O
Transport

Available Tools (33)

TOP_P

`0.95`

Hunyuan

`HUNYUAN_API_KEY`

MULTI_CROP_MAX_TILES

`5`

MULTI_CROP

`true`

ENABLE_THINKING

`true`

MODEL_PROVIDER

`zhipu`

Zhipu

`ZHIPU_API_KEY`

MODEL_NAME

自动选择

SiliconFlow

`SILICONFLOW_API_KEY`

Qwen

`DASHSCOPE_API_KEY`

TEMPERATURE

`0.7`

BASE_VISION_PROMPT

内置默认值

Volcengine

`VOLCENGINE_API_KEY`

MAX_TOKENS

`8192`

CUSTOM_BASE_URL

OpenAI 兼容 API 地址（默认 `https://api.minimaxi.com/v1`）

CUSTOM_MODEL_NAME

模型名称（默认 `MiniMax-M3`）

CUSTOM_API_KEY

API Key

zhipu

`ZHIPU_API_KEY`

siliconflow

`SILICONFLOW_API_KEY`

qwen

`DASHSCOPE_API_KEY`

volcengine

`VOLCENGINE_API_KEY`

hunyuan

`HUNYUAN_API_KEY`

custom

`CUSTOM_API_KEY` + `CUSTOM_BASE_URL` + `CUSTOM_MODEL_NAME`

image_source

是

prompt

是

task_type

否

INCLUDE_META

`false`

LUMA_DEBUG

关闭

MCP_TRANSPORT

`stdio`

MCP_HTTP_HOST

`0.0.0.0`

MCP_HTTP_PORT

`3000`

MCP_HTTP_TOKEN

空（不鉴权）

describe

`task_type` 行为： - 省略或 `auto`（默认）：与旧版一致，按 prompt 启发式路由 - `ocr`：文字提取，默认单图高保真（关闭 multi-crop） - `ui` / `debug`：界面结构 / 报错截图，倾向文本保真 - `describe`：简短描述 示例： ```typescript image_understand({ image_source: "./screenshot.png", prompt: "分析这个页面的布局和主要组件结构", task_type: "ui", }); image_understand({ image_source: "./code-error.png", prompt: "这段代码为什么报错？请给出修复建议", // task_type 可省略，行为与旧版兼容 }); image_understand({ image_source: "https://example.com/ui.png", prompt: "找出这个界面的可用性问题", }); ``` ### 使用建议 - 非视觉模型需要明确提示调用 MCP 工具 - 代码截图、OCR、长图、表格这类文本密集图片会自动启用更保真的处理方式 - 大图会按配置自动生成原图加裁剪图，提高细节理解能力 - 需要排查耗时/裁剪数时设 `INCLUDE_META=true` 或 `LUMA_DEBUG=1`，结果末尾会附 `luma_meta` ## 环境变量 ### 通用配置

Browse all MCP servers