Services

Service Comparison

Service	Lang/Runtime	Framework	Port (Main)	Port (Dev)	Scale	Purpose
Frontend	TypeScript	React 19 + Vite 7	3000	3001	Main + Dev	User-facing dashboard
Backend	Java 17	Spring Boot 3	8080	8081	Main + Dev	REST API, JWT auth, workspace management
Analyzer	TypeScript	Node.js/Express	9000	9001	Main + Dev	AI log analysis (Groq, 27 strategies)
Semantic Search	TypeScript	Node.js/Express	9100	—	Main only	Neural log search (sentence-transformers)
AI Service	TypeScript	Node.js/Express	9300	—	Main only	Conversational AI with tool calling (Groq)
MCP Server	Python 3.10	FastAPI + uvicorn	9500	—	Main only	MCP-compatible compliance tools
Agent Runner	TypeScript	Node.js (proposed)	9400	—	Main only	Background agentic scheduler
Webhook Server	JavaScript	Node.js built-ins	9999	—	Single	GitHub auto-deploy
OpenSearch	Java 21	OpenSearch 3.4	9200	—	Single	Search engine + ML plugins
OS Dashboards	Node.js	OpenSearch-Dashboards	5601	5603	Single	Visualization UI
Data Prepper	Java	Docker container	21890/21891	—	Single	OTEL trace/log ingestion

Descriptions

Frontend

React 19 + Vite 7 SaaS dashboard with:

• Redux Toolkit for global state
• Tailwind CSS with dark/light themes
• ECharts for analytics charts
• WebSocket-based scan progress tracking
• Lazy-loaded heavy components (CortexPanel, ECharts: ~1.1MB)
• 16 optimized JS chunks (down from 125)

Backend

Spring Boot 3 REST API with:

• JWT-based authentication
• Multi-tenant workspace isolation
• Compliance scan management
• Integration with OpenSearch for log data
• PostgreSQL (or H2 for dev) persistence

Analyzer

Node.js/TypeScript AI analysis engine with:

• 27 analysis strategies via Groq (llama-3.3-70b-versatile)
• Log pattern recognition and key findings extraction
• Swagger UI at /api-docs
• Health endpoint at /health
• .env loaded per-branch for isolated configuration

Semantic Search

Neural search API that:

• Wraps OpenSearch k-NN with sentence-transformers (all-MiniLM-L6-v2, 384 dims)
• Exposes /api/v1/search/all (neural + keyword combined results)
• Exposes /api/v1/indexes (all OpenSearch indexes with has_knn flag)
• Auto-detects semantic vs non-semantic indexes
• Main-branch only — no dev instance

AI Service

Conversational AI interface with:

• SSE streaming chat via /api/v1/chat
• Multi-turn tool calling via AgentLoop
• 4 tool categories: semantic search, OpenSearch queries, Backend API, Analyzer
• SQLite for conversation persistence
• WebSocket for typing/presence indicators
• Swagger UI at /api-docs

MCP Server

Python/uvicorn service that:

• Exposes compliance scanning tools via HTTP+SSE MCP transport
• Compatible with Claude Desktop, Cursor, and any MCP client
• Health endpoint at /health
• SSE endpoint at /sse
• Production: https://mcp.logrite.ai

Agent Runner (Proposed — Port 9400)

New Node.js/TypeScript service that:

• Polls Backend API for scheduled agents
• Registers cron jobs for each active agent
• On fire: builds context, calls AI Service /chat, stores execution log
• Evaluates alert conditions and notifies Discord
• See AI Agents: Execution Loop

Webhook Server

Minimal Node.js service using only built-ins:

• Listens for GitHub push events on port 9999
• Verifies HMAC-SHA256 signatures
• Receives the push event; the deploy watcher builds and deploys through the console
• Config from config/webhook.json
• No npm dependencies

OpenSearch

OpenSearch 3.4.0 built from source with plugins:

• opensearch-ml — ML Commons framework
• opensearch-knn — Vector similarity search
• opensearch-anomaly-detection — ML-based anomaly detection
• opensearch-alerting — Monitor + notification engine
• opensearch-notifications — Discord webhook delivery