How pasittum/rag_ai works
Self-hosted AI Package is a Docker Compose template for running a local AI and low-code environment. The main user workflow is to start the selected service profile, then chat through Open WebUI; the included n8n Pipe sends the latest message to an n8n webhook and returns its response. The repository also includes RAG and agent workflow assets, search, vector storage, and supporting services. The README documents the broader stack, but most service definitions and workflow JSON were not sampled, so their internal connections are not inferred.
Main entry points: docker-compose.yml Compose stack · n8n_pipe.py n8n Pipe · start_services.py Profile launcher · Web Search + n8n Agent Chatflow.json Flowise chatflow
Components 16
Startup
- Profile launcher
start_services.py - Compose stack
docker-compose.yml
Chat and agents
- Open WebUI
docker-compose.yml - n8n Pipe
n8n_pipe.py - n8n automation
docker-compose.yml - RAG workflows
n8n/backup/workflows - RAG agent workflow
RAG.json - Flowise chatflow
flowise/Web Search + n8n Agent Chatflow.json - Ollama models
docker-compose.yml
Data and retrieval
- Supabase
supabase - Qdrant vector store
docker-compose.yml
Supporting services
- SearXNG search
searxng/settings-base.yml - Neo4j
docker-compose.yml - Langfuse
docker-compose.yml - Caddy proxy
Caddyfile
Other components
- User
Connections 16
- User → Open WebUI: chats in
README.md - Profile launcher → Compose stack: starts profile
start_services.py - Compose stack → Open WebUI: runs service
README.md - Compose stack → n8n automation: runs service
README.md - Compose stack → Ollama models: runs service
README.md - Compose stack → Supabase: runs service
README.md - Compose stack → Qdrant vector store: runs service
README.md - Compose stack → Neo4j: runs service
README.md - Compose stack → Langfuse: runs service
README.md - Compose stack → SearXNG search: runs service
README.md - Compose stack → Caddy proxy: runs service
README.md - Open WebUI → n8n Pipe: invokes pipe
- n8n Pipe → n8n automation: posts webhook
n8n_pipe.py - n8n automation → RAG workflows: runs workflows
- n8n automation → RAG agent workflow: runs workflow
- Flowise chatflow → n8n automation: uses agent