Skip to main content
Glama

Solid Multi-Tenant DevOps MCP Server

SUMMARY.mdβ€’9.73 kB
# Solid Multi-Tenant DevOps MCP Server - Summary ## βœ… What Was Created A **standalone MCP server** for AI-first DevOps monitoring and management of your multi-tenant Solid SaaS platform. ### Location ``` /Users/adamcampbell/Desktop/Solid/solid-mcp-server/ ``` ### Files Created - `src/index.js` - Main MCP server implementation (19 tools) - `package.json` - Node.js dependencies - `.env` - Service endpoint configuration - `README.md` - Full documentation - `INSTALL.md` - Installation guide - `claude-desktop-config.json` - Claude Desktop configuration - `test-connection.sh` - Connection test script ## 🎯 What It Does This MCP server connects Claude Desktop to your Solid platform, allowing you to: ### Monitor Thousands of Tenants - List all tenant companies (with pagination for scale) - Search tenants by name, email, or ID - Get detailed tenant information - Bulk health checks for multiple tenants - Platform-wide analytics across all tenants ### Per-Tenant AI Agent Monitoring Each tenant has 12 AI agents: 1. **ADA** - Orchestrator (VP of AI) 2. **Sarah** - Customer Service Manager 3. **Jake** - Inventory Manager 4. **Devon** - Strategy 5. **Marcus** - Marketing Manager 6. **Alex** - Finance Manager 7. **Jordan** - Operations Manager 8. **Maya** - Brand Manager 9. **Riley** - Graphic Designer 10. **Casey** - Video Creator 11. **Taylor** - Affiliate Manager 12. **Morgan** - DevOps Engineer You can: - Monitor all agents per tenant - View inter-agent communication - Track agent performance metrics - Monitor token usage and costs (2.5x markup model) ### System Health Monitoring - Check all services (Backend, Agents, Docker) - Monitor Docker containers (PostgreSQL, Redis, Backend) - View real-time performance metrics - Identify slow API endpoints - Track errors across all tenants ### Error Management - List recent errors (all tenants or specific) - Filter by severity (critical, error, warning, info) - Get detailed error information with stack traces - Mark errors as resolved with notes ### Tenant Provisioning - Provision new tenant companies - Configure subscription plans (starter, professional, enterprise) - Enable demo mode with sample data - Automated setup workflow ## πŸ”§ Current System Status ### βœ… All Services Running - **Backend**: http://localhost:8090 (API server) - **Agents**: http://localhost:8091 (AI agent execution) - **Super Admin**: http://localhost:8080 (Platform management UI) - **Frontend**: http://localhost:3000 (Customer-facing) - **CMS**: http://localhost:3001 (Public CMS) ### βœ… Database Services - **PostgreSQL**: solid-postgres container (running) - **Redis**: solid-redis container (running) ### βœ… Current Tenant Data - **1 tenant company** (Test Company) - **12 AI agents** active for company ID 1 - All agents status: active ## πŸ“‹ 19 MCP Tools Available ### Tenant Management (5 tools) 1. `list_all_tenants` - List all tenants with pagination 2. `get_tenant_details` - Get specific tenant info 3. `get_platform_analytics` - Platform-wide metrics 4. `search_tenants` - Search by name/email/ID 5. `bulk_check_tenant_health` - Health check multiple tenants ### System Health (2 tools) 6. `check_all_services_health` - Check all services 7. `check_docker_containers` - Docker container status ### Per-Tenant Monitoring (4 tools) 8. `get_tenant_agents` - All AI agents for a tenant 9. `get_tenant_agent_messages` - Inter-agent communication 10. `get_tenant_products` - Products/inventory 11. `get_tenant_token_usage` - AI token usage & billing ### Error Monitoring (3 tools) 12. `list_recent_errors` - Recent errors (filtered) 13. `get_error_details` - Detailed error info 14. `resolve_error` - Mark error as resolved ### Performance Monitoring (2 tools) 15. `get_performance_summary` - System performance metrics 16. `get_slow_endpoints` - Find slow API endpoints ### Tenant Provisioning (1 tool) 17. `provision_new_tenant` - Create new tenant ### Database Operations (1 tool) 18. `execute_raw_query` - Raw SQL queries (READ-ONLY) ### Bulk Operations (1 tool) 19. `bulk_check_tenant_health` - Batch health checks ## πŸš€ Next Steps ### 1. Install in Claude Desktop ```bash # Copy config to Claude Desktop mkdir -p ~/Library/Application\ Support/Claude cp claude-desktop-config.json ~/Library/Application\ Support/Claude/claude_desktop_config.json # Restart Claude Desktop ``` Or follow the detailed instructions in `INSTALL.md`. ### 2. Test the Connection Open Claude Desktop and try: ``` Show me all tenants in the system ``` ``` Check the health of all services ``` ``` Show me all AI agents for company 1 ``` ### 3. Example Use Cases #### Morning Health Check ``` Check all services health, then show me any critical errors from last night, and identify any slow endpoints ``` #### Tenant Monitoring ``` Show me company 1's details, list their AI agents with status, and check their token usage ``` #### Bulk Operations ``` Check the health of companies 1, 2, and 3 ``` #### Performance Investigation ``` Find endpoints slower than 2 seconds and show me recent errors ``` ## 🎨 Super Admin Dashboard Your Super Admin dashboard at http://localhost:8080/admin provides the visual interface for: - Viewing all tenants - Monitoring system health - Managing platform configuration - Viewing error logs - Managing subscriptions The MCP server complements this by allowing **natural language queries** through Claude Desktop. ## πŸ” Architecture Notes ### Multi-Tenant Design - Each tenant is a `Company` with unique `company_id` - All data scoped per company (agents, products, users, orders) - Platform supports thousands of tenant instances - Each tenant is isolated from others ### Service Architecture ``` β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ Claude Desktop (You) β”‚ β”‚ "Show me all tenants and their health" β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ ↓ stdio β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ Solid MCP Server (Node.js) β”‚ β”‚ Translates natural language β†’ API callsβ”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ ↓ HTTP β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ Backend API (FastAPI) β”‚ β”‚ http://localhost:8090 β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ ↓ β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ PostgreSQL + Redis β”‚ β”‚ (Docker containers) β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ ``` ## πŸ“Š Scaling Expectations This MCP server is designed to handle: - **Thousands of tenant companies** - **12 AI agents per tenant** = 12,000+ total agents - **Millions of inter-agent messages** - **Bulk operations** for monitoring hundreds of tenants at once - **Real-time performance monitoring** across all tenants ## πŸ› οΈ Troubleshooting ### Run Connection Test ```bash cd /Users/adamcampbell/Desktop/Solid/solid-mcp-server ./test-connection.sh ``` ### Start All Services ```bash cd /Users/adamcampbell/Desktop/Solid ./start-all.sh ``` ### View Logs ```bash # Backend logs tail -f /Users/adamcampbell/Desktop/Solid/logs/backend.log # Agent logs tail -f /Users/adamcampbell/Desktop/Solid/logs/agents.log # Docker logs docker-compose logs -f ``` ### Check Docker Containers ```bash docker ps | grep solid ``` ## 🎯 Key Differences from Previous Attempt ### ❌ What I Did NOT Do - Did NOT integrate MCP into your existing backend server - Did NOT modify your backend code - Did NOT touch your existing servers at all ### βœ… What I DID Do - Created a **completely separate, standalone** MCP server - Connects to your existing APIs as a **client** - Runs independently in Claude Desktop - Calls your existing endpoints (no backend changes needed) ## πŸ“ Technical Details ### Dependencies - `@modelcontextprotocol/sdk` - MCP protocol implementation - `node-fetch` - HTTP requests to your APIs ### Configuration All endpoints configured in `.env`: ```env BACKEND_URL=http://localhost:8090 SUPER_ADMIN_URL=http://localhost:8080 AGENTS_URL=http://localhost:8091 ``` ### No Authentication (Yet) Currently connects to localhost endpoints without auth. When you deploy to production, you'll want to: - Add API authentication tokens - Use environment variables for secrets - Implement role-based access control ## πŸŽ‰ Success Criteria You know it's working when: 1. βœ… `./test-connection.sh` shows all services healthy 2. βœ… Claude Desktop shows "solid-devops" in MCP servers list 3. βœ… You can ask Claude "Show me all tenants" and get results 4. βœ… You can monitor thousands of companies through natural language ## πŸ“š Documentation - `README.md` - Full feature documentation - `INSTALL.md` - Step-by-step installation guide - `SUMMARY.md` - This file (overview) --- **Status**: βœ… Ready for multi-tenant monitoring **Version**: 1.0.0 **Built for**: AI-first DevOps monitoring of Solid SaaS platform

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Adam-Camp-King/solid-mcp-server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server