# homeai-kit-test

A full local AI stack for personal homelab use. Bundles Ollama (local LLM inference), Open WebUI (chat interface), ChromaDB (vector database), SearXNG (private search), and n8n (workflow automation) into a single Docker Compose deployment.

This is a test/evaluation deployment of the **HomeAI Kit** stack on the Linux server.

## Stack

### Core Services (always started)

| Service | Container | Port | Description |
|---------|-----------|------|-------------|
| Ollama | `homeai-ollama` | `11435` | Local LLM inference server |
| Open WebUI | `homeai-open-webui` | `3000` | Chat UI connected to Ollama |
| ChromaDB | `homeai-chromadb` | `8100` | Vector database (token-auth enabled) |
| SearXNG | `homeai-searxng` | `8080` | Private meta-search engine |
| PostgreSQL | `homeai-postgres` | `5432` | Database backend for n8n |
| n8n | `homeai-n8n` | `5678` | Workflow automation |

### Optional Services (`--profile extras`)

| Service | Container | Port | Description |
|---------|-----------|------|-------------|
| LibreTranslate | `homeai-libretranslate` | `5000` | ML translation API |
| Redis | `homeai-redis` | `6379` | Caching layer |
| ChromaDB Admin | `homeai-chromadb-admin` | `3002` | Vector DB admin UI |

## Running

```bash
# Core stack only
docker compose up -d

# Core + optional extras
docker compose --profile extras up -d

# Stop everything
docker compose down
```

## Setup

```bash
cp .env.example .env
# Edit .env — at minimum change POSTGRES_PASSWORD and CHROMA_SERVER_AUTHN_CREDENTIALS
```

After startup, pull an LLM model:
```bash
docker exec homeai-ollama ollama pull llama3.2:3b
# Or use the helper script:
./scripts/pull-models.sh llama3.2:3b
```

Then open `http://localhost:3000` to create your Open WebUI admin account.

## Configuration

All ports and credentials are set in `.env`. Key settings:

| Variable | Default | Description |
|----------|---------|-------------|
| `OLLAMA_PORT` | `11435` | Ollama API port |
| `WEBUI_PORT` | `3000` | Open WebUI port |
| `POSTGRES_PASSWORD` | *(required)* | PostgreSQL password |
| `N8N_PORT` | `5678` | n8n workflow UI |
| `CHROMA_PORT` | `8100` | ChromaDB port |
| `CHROMA_SERVER_AUTHN_CREDENTIALS` | *(required)* | ChromaDB auth token |
| `SEARXNG_PORT` | `8080` | SearXNG port |

## Stacks

The `stacks/` directory contains modular sub-stacks for selective deployment:

| Stack | Description |
|-------|-------------|
| `stacks/llm/` | Ollama + Open WebUI only |
| `stacks/rag/` | ChromaDB + SearXNG for RAG pipelines |
| `stacks/automation/` | n8n + PostgreSQL |
| `stacks/extras/` | LibreTranslate, Redis, ChromaDB Admin |

## SearXNG Config

SearXNG settings are in `config/searxng/settings.yml` and `config/searxng/limiter.toml`. These are mounted read-only into the container.

## Data Persistence

All service data is stored in named Docker volumes:

| Volume | Service |
|--------|---------|
| `ollama_data` | Downloaded LLM models |
| `openwebui_data` | Chat history, users |
| `chromadb_data` | Vector embeddings |
| `postgres_data` | n8n workflows |
| `n8n_data` | n8n credentials |
| `libretranslate_data` | Translation models |

## Deployment

Runs on `192.168.0.241` (albert-MacBookPro Linux server).

**Docker Compose:** `/opt/docker/homeai-kit-test/docker-compose.yml`

| Service | URL |
|---------|-----|
| Open WebUI | `http://192.168.0.241:3000` |
| n8n | `http://192.168.0.241:5678` |
| Ollama API | `http://192.168.0.241:11435` |
| SearXNG | `http://192.168.0.241:8080` |
| ChromaDB | `http://192.168.0.241:8100` |

> **Note:** This is a separate Ollama instance from the one used by ScraperControl (which runs as a systemd service on port `11434`). This stack uses port `11435` to avoid conflict.