PlebChat

initialize

# setup .env vars first
cp .env.example .env

mkdir ./volumes/backend_data
mkdir ./volumes/pipelines_dir

Production setup

docker compose -f docker-compose.yml --env-file .env up -d

Development setup

docker compose -f docker-compose-dev.yml --env-file .env up -d

check logs

https://containrrr.dev/watchtower/arguments/

docker logs watchtower

---

System Overview / Architecture Diagram

graph TD
    A[Open WebUI - Docker]
    B[Pipelines - Python Server]
    Cl[`cloudflared`]
    D[PlebChat Graph - Python Server]

    A -->|Web request| B
    Cl -->|NAT address translation| A
    B -->|user query| D
    D -->|verify user status| G[PlebChat DB - Python Server]
    G -->|user balance| D
    G -->|query database| H[MongoDB Server Replica Mode]
    I[Admin Panel] -->|monitors| G
    D -->|LLM completion| J[Ollama Server]

Loading

Open WebUI

See website, repository, documentation and Setup instructions

This is an open-source repository used as the "frontend."

pipeline

See repository and setup instructions

Think of Open WebUI pipelines as extensions. This repository stands up an OpenAI-compatible API that is used to invoke our LangGraph server.

LangGraph agent

See repository and setup instructions

This is our LangGraph agent.

User database

See respository and setup instructions

This database tracks token balances and usage history of registered users.

This includes a FastAPI server which communicates with a MongoDB database in replication mode.

An admin panel written with Streamlit allows for local maintenance.