Skip to main content
Glama
README.md
# 🧠 VibeLogic MCP

**为 Cursor 打造的确定性架构引擎与逻辑死锁协议**

当项目复杂度提升时,AI 编程往往会陷入“幻觉螺旋”。VibeLogic MCP 并不是另一个代码生成器,而是一个处于“人类意图”与“代码执行”之间的管控层。它强制 AI 在动刀改代码前,先通过可视化蓝图进行审计。

## ✨ 核心能力

- **可视化架构审计 (Visual Audit)**:强制 AI 提取项目拓扑,通过 Mermaid 蓝图对齐逻辑,确保 100% 意图确定性。
- **逻辑死锁协议 (Security Lock)**:独家封装的注意力劫持协议,强制 AI 在“图纸确认”前保持原地待命,阻断越权重构。
- **跨文件深度追踪**:自动识别多文件关联变更,生成结构化的施工清单 (Action Plan)。
- **资产化存档**:所有架构决策自动导出为 `.md` 文件,永久沉淀为项目的架构演进日志。

## 🚀 快速开始

##  1. 安装
确保你已安装 Node.js,然后在 Cursor 中直接运行:
```bash
npx vibelogic-mcp
```

## 2. 在 Cursor 中配置
打开 Cursor Settings -> Features -> MCP。
点击 + Add New MCP Server。
设置如下:
```
{
  "mcpServers": {
    "VibeLogic": {
      "command": "npx.cmd",
      "args": [
        "-y",
        "vibelogic-mcp@latest"
      ]
    }
  }
}
```

## 💡 常用咒语 (Prompts)

"帮我分析下这个功能的逻辑,我想看图。"
"我要重构登录模块的校验流程,先出图和施工单。"
"基于当前代码,审计一下新增 API 对现有架构的影响。"

## 🔒 隐私与安全

BYOK (自带 Key):VibeLogic 运行在你的本地环境,不存储任何代码,完全复用你 Cursor 选定的模型能力。
拒绝复杂化:仅在涉及核心逻辑变动时激活,不干扰常规的 UI 调整和 Bug 修复。

TDQS

A3.6/5.0

Scored across 1 tool

Disambiguation5/5

With only one tool, there is no risk of confusion between tools.

Naming Consistency5/5

The single tool uses a verb_noun pattern (get_logic_blueprint), and with only one tool, consistency is not an issue.

Tool Count2/5

One tool is too few for a server named 'vibelogic', which suggests a need for multiple logic-related operations. The tool is also highly restricted in usage, further limiting its utility.

Completeness1/5

A single tool cannot cover the expected lifecycle or operations for logic manipulation. The tool itself warns against many common tasks, leaving obvious gaps.

Maintenance

ActivityInactive
ResponsivenessNo issues