Claude MCP Server Ecosystem
A production-ready MCP server ecosystem providing Claude AI with 150+ specialized tools across enhanced memory, data analytics, security, design, and infrastructure domains with PostgreSQL, Redis, Qdrant, and Docker orchestration.
README
🚀 Claude MCP Server Ecosystem
🚨 CRITICAL: Two Issues That Keep Causing Problems
1. GLOBAL CONFIG REQUIRED
- MCP configs MUST be global (
~/.claude/claude_code_config.json), NOT local/project-specific - Local configs are lost when changing directories
2. STDIO TRANSPORT REQUIRED
- ALL MCP servers MUST use STDIO communication, NOT HTTP/ports
- Claude Code only supports STDIO transport
Quick Fix for MCP Issues:
# If MCP tools not working in Claude Code:
exit # Exit current session
claude --mcp-config ~/.claude/claude_code_config.json # Restart with global config
Status: ✅ OPTIMIZED SYSTEM ARCHITECTURE IMPLEMENTED (2025-05-26)
Servers: 13 configured (optimized from 17 → 13 via consolidation)
Tools: 150+ available including consolidated data analytics and enhanced memory optimization
Memory: 58% reduction in memory overhead through server consolidation
⚡ INSTANT START (60 SECONDS)
1. Verify System Status:
bash verify-system-status.sh
2. Start MCP Ecosystem:
Option A - PM2 (Recommended for Production):
./start-mcp-pm2.sh
Option B - CLI (Legacy):
./claude-mcp-setup start
Expected: "All 11 servers ready, 0 failed"
3. Test in Claude Code:
/mcp
Expected: 11 connected servers listed (including enhanced-memory)
📚 DOCUMENTATION
🚀 START HERE
- ⚡ IMMEDIATE START:
START_HERE_IMMEDIATE.md- Where you are right now, what to do next - 💻 ENVIRONMENT STATUS:
DEVELOPMENT_ENVIRONMENT_STATUS.md- Current system state, process status, validation - ⌨️ COMMAND REFERENCE:
COMMAND_REFERENCE_IMMEDIATE.md- Copy-paste ready commands for all tasks
📖 COMPLETE DOCUMENTATION
- ⚡ Latest Session:
SESSION_2025-05-26_SYSTEM_OPTIMIZATION.md- Server consolidation, memory optimization, and performance improvements - 📊 Optimization Summary:
OPTIMIZATION_SUMMARY_2025-05-26.md- Executive summary of optimization results - 📖 Current State:
CURRENT_WORKING_STATE.md- Updated with session 2025-05-26 optimizations - 🧠 Previous Session:
SESSION_2025-05-25_ENHANCED_MEMORY_IMPLEMENTATION.md- Enhanced memory system + MCP fixes - 📖 Complete Guide:
README_AUTHORITATIVE_2025-05-24.md - 📁 Historical Docs:
docs/archive/(for reference only)
🎯 WHAT THIS PROJECT IS
A production-ready MCP server ecosystem that gives Claude AI 150+ specialized tools across:
- 📊 Consolidated Data Analytics (40+ tools): Unified server combining pipelines, warehousing, governance, ML deployment, real-time analytics
- 🧠 Enhanced Memory (3 tools): 6 optimization techniques including context compression, hierarchical memory, semantic chunking
- 🔐 Security (6 tools): Vulnerability scanning, compliance checking
- 🎨 Design (8 tools): UI analysis, accessibility, design systems
- 💾 Memory Management (5 tools): Persistent storage, context management
- ⚡ Optimization (5 tools): Performance profiling, bottleneck analysis
- 🏗️ Infrastructure (85+ tools): Database, Docker, configuration management
Key Optimization (2025-05-26): Consolidated 5 separate data analytics servers into 1 unified server, reducing memory usage by ~350MB while maintaining full functionality.
Infrastructure: PostgreSQL + Redis + Qdrant + Docker orchestration
⚠️ IMPORTANT
If you see any issues:
- Run
bash verify-system-status.shfor diagnosis - Check
README_AUTHORITATIVE_2025-05-24.mdfor troubleshooting - Ignore other markdown files (they contain outdated information)
Last Verified: 2025-05-24
Root Cause Fixed: claude-mcp-wrapper.sh environment variable configuration
Result: All 10 servers working with 149 tools available
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。