beerman
MCP server for long-term project memory management, allowing AI agents to store and semantically search project milestones, decisions, and architecture specs via Supermemory vector store.
README
🍺 BeerMan - Long-Term Project Memory Engine
BeerMan is a professional pair-programming memory system that acts as a continuous intelligence bridge between your IDE agents (Anti-Gravity, Claude Code, Cursor) and your planning sessions. It leverages Supermemory as a centralized, canonical vector store for all architectural guidelines, design decisions, code watchers, and context captures.
┌──────────────────────────────────────────────────────────┐
│ Planning & Session │
│ (ChatGPT / Claude Web Chats / Scrolling Screenshots) │
└────────────────────────────┬─────────────────────────────┘
│ (Browser Sidebar Extension)
▼
┌──────────────────────────────┐
│ Supermemory Vector Store │
└──────────────┬───────────────┘
│ (Memory Bridge / MCP API)
▼
┌──────────────────────────────────────────────────────────┐
│ Development Workspace │
│ (Anti-Gravity / Claude Code / Cursor / VS Code) │
└──────────────────────────────────────────────────────────┘
⚡ Quick Start (One-Liner)
Run the following command in your terminal to instantly clone, install dependencies, compile TypeScript packages, and start the development servers:
git clone https://github.com/nitineeen/beerman.git && cd beerman && npm install && npm run build && npm run dev
Once running, your local dashboard will be hosted at http://localhost:2414.
🏗️ System Architecture & Monorepo Layout
BeerMan is built as a TypeScript monorepo where components have single, isolated responsibilities:
project-beerman/
├── apps/
│ ├── extension/ # Unpacked Manifest V3 Chrome Sidebar Extension
│ └── dashboard/ # Next.js workspace management dashboard & setup guide
├── packages/
│ ├── core/ # Shared project type definitions, categories, and models
│ ├── memory/ # Supermemory client SDK for ingestion and semantic searches
│ └── mcp-server/ # Model Context Protocol (MCP) server integration
└── .agents/
└── skills/ # Structured agent guidelines (YAML + Markdown format)
1. Chrome Extension Sidebar (apps/extension)
- Slides out on the right viewport, automatically resizing and rendering target webpages responsively.
- Direct Supermemory integration for live project creation, renaming, and document deletion.
- Features sequential upload queueing with robust error handling and duplicate prevention.
- Seamlessly captures raw text selections, full-page screenshots, files, and chat logs.
2. BeerMan Dashboard (apps/dashboard)
- Next.js web application that displays your living docs, project indexing details, and setup guides.
3. Model Context Protocol Server (packages/mcp-server)
- Exposes tools to your AI agent so it can interact with your memory directly:
list_projects: Lists registered project indices.open_project(projectId): Selects active context.store_memory(content, category, tags): Commits structured milestones, decisions, architecture specs, or roadmaps.search_memory(query): Performs vector/semantic search on project memory.
🚀 Setup & Installation
Step 1: Install Dependencies & Build
Compile the TypeScript packages and Next.js applications:
npm install
npm run build
Step 2: Configure the MCP Server in your IDE
Add the BeerMan MCP server configuration to link it with your agent. Ensure you replace /absolute/path/to/project-beerman with your actual directory path.
For Anti-Gravity or Claude Code:
Add this configuration block to your global configurations at ~/.gemini/config/mcp_config.json or ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"beerman": {
"command": "node",
"args": ["/absolute/path/to/project-beerman/packages/mcp-server/dist/index.js"],
"env": {
"SUPERMEMORY_API_KEY": "your_supermemory_api_key_here"
}
}
}
}
For Cursor:
- Navigate to Settings > Features > MCP.
- Click + Add New MCP Server.
- Fill out the fields:
- Name:
beerman - Type:
command - Command:
node /absolute/path/to/project-beerman/packages/mcp-server/dist/index.js
- Name:
- Under Environment Variables, add
SUPERMEMORY_API_KEYwith your API key.
Step 3: Install the Chrome Sidebar Extension
- Open Google Chrome and go to
chrome://extensions/. - Toggle Developer mode on (top-right corner).
- Click Load unpacked (top-left corner).
- Select the directory:
project-beerman/apps/extension. - Click the extension icon in your toolbar to open the right-side panel.
- Enter your Supermemory API Key, and manage your projects dynamically.
🛠️ Deploying & Sharing
1. Deploying the Dashboard (Vercel)
The dashboard can be instantly deployed to Vercel:
- Connect this repository to your Vercel Account.
- Set the Root Directory to
apps/dashboard. - Vercel will automatically compile the Next.js bundle and deploy it.
2. Sharing the Extension
- Unpacked: Zip the
apps/extensionfolder and distribute it to your team to load via Developer Mode. - Chrome Web Store: Package the extension and upload it to the Chrome Developer Dashboard to publish it.
3. MCP Server Cloud Deployment
You can host the MCP server on Railway, Fly.io, or Render, exposing it via SSE (Server-Sent Events) or standard stdio transports, allowing team members to connect to the same remote memory.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。