luki-api-gateway

This repository is archived. Active development continues in a private repository. This public version reflects the architecture from the ReMeLife integration era and is no longer maintained.

Single entry point for clients to access LUKi AI services: chat, memory, cognitive modules, and reporting.

What It Does

• Routes requests to downstream microservices (core agent, memory, cognitive, engagement, reporting)
• Authenticates via JWT and API keys
• Rate-limits per user with Redis (in-memory fallback)
• Streams chat responses via SSE
• Proxies ELR memory operations and module endpoints
• Adds correlation IDs to all requests for tracing

Stack

• Framework: FastAPI + Uvicorn
• Auth: python-jose (JWT), custom middleware
• Rate Limiting: Custom Redis-based limiter with in-memory fallback
• HTTP Clients: httpx (async) to downstream services
• Deployment: Docker on Railway

Structure

luki_api/
├── main.py                  # FastAPI app, middleware registration
├── config.py                # Env vars, service URLs, rate limit settings
├── middleware/
│   ├── auth.py              # JWT + API key authentication
│   ├── rate_limit.py        # Redis + in-memory rate limiting
│   ├── metrics.py           # Prometheus metrics collection
│   ├── cache.py             # Response cache layer
│   ├── circuit_breaker.py   # Circuit breaker for downstream calls
│   ├── correlation.py       # Request correlation IDs
│   └── logging.py           # Structured request logging
├── routes/
│   ├── chat.py              # POST /api/chat/{user_id}/stream (SSE)
│   ├── conversation.py      # Conversation management
│   ├── conversations.py     # Conversation listing
│   ├── elr.py               # ELR memory proxy
│   ├── memories.py          # Memory operations
│   ├── uploads.py           # File upload handling
│   ├── health.py            # GET /health
│   ├── metrics.py           # GET /metrics
│   ├── cognitive.py         # Cognitive module proxy
│   └── wallet.py            # Wallet verification proxy
├── clients/
│   ├── agent_client.py      # Core agent HTTP client (SSE streaming)
│   ├── memory_service.py    # Memory service HTTP client
│   ├── base_client.py       # Shared HTTP client base
│   ├── security_service.py  # Security service HTTP client
│   └── wallet_client.py     # Solana wallet verification client
└── monitoring/
    └── health_monitor.py    # Downstream service health tracking

Prerequisites

This gateway proxies requests to other LUKi microservices. You need running instances of:

• luki-core-agent — AI chat engine (required for /api/chat)
• luki-memory-service — ELR memory storage (required for /api/elr)
• Redis — rate limiting (optional, falls back to in-memory)

Setup

git clone git@github.com:ReMeLife/luki-api-gateway.git
cd luki-api-gateway
python -m venv venv && source venv/bin/activate
pip install -r requirements.txt
uvicorn luki_api.main:app --reload --port 8081

Environment variables (set in .env or export):

export SUPABASE_URL=https://your-project.supabase.co
export SUPABASE_JWT_SECRET=your-jwt-secret
export LUKI_CORE_AGENT_URL=http://localhost:9000
export LUKI_MEMORY_SERVICE_URL=http://localhost:8002
export LUKI_COGNITIVE_SERVICE_URL=http://localhost:8101
export REDIS_URL=redis://localhost:6379

Key Endpoints

Method	Path	Description
POST	/api/chat/{user_id}/stream	SSE streaming chat
POST	/api/elr/ingest_text	Ingest text to memory
POST	/api/elr/search	Semantic memory search
GET	/health	Service health check
GET	/metrics	Prometheus metrics

License

Apache License 2.0. See LICENSE.