QUOT — Quary Unified Observability Tool

A modular Retrieval-Augmented Generation (RAG) system supporting multiple LLM providers. Built with LangChain, ChromaDB, and Streamlit, with a FastAPI REST API and a React dashboard.

Features

Multi-Provider Support — OpenAI, Anthropic, Google, HuggingFace, IBM Watsonx, or all at once via multi/
Multiple Interfaces — Streamlit UI per provider, FastAPI REST API, and Chainlit chat
XML-Structured Retrieval — Retrieved chunks in <document> tags; full context in <context> tags for cleaner LLM delineation
Unified Logging — Single loggers/ module writes JSONL events; optionally forwards to Weights & Biases or MLflow
React Dashboard — Real-time view of queries, sessions, errors, and response times (dashboard/)

Supported Providers

Provider	LLM Models	Embeddings	Local Option
OpenAI	GPT-4o, GPT-4, GPT-3.5	text-embedding-3-small	No
Anthropic	Claude 3.5 Sonnet/Haiku, Claude 3 Opus	Google or OpenAI (via `EMBEDDING_PROVIDER`)	No
Google	Gemini 2.0 Flash, Gemini 1.5 Pro	embedding-001	No
HuggingFace	Mistral-7B, Llama-2, TinyLlama	sentence-transformers	Yes
IBM Watsonx	Mixtral-8x7B, Granite, Llama-3	HuggingFace (local)	No

Project Structure

QUOT/
├── anthropic/               # Anthropic Claude provider
│   ├── app.py               # Streamlit UI
│   ├── qa_bot.py            # RetrievalQA chain (Claude)
│   ├── embeddings.py        # Google embeddings
│   ├── embeddings_openai.py # OpenAI embeddings (alternative)
│   ├── retriever.py
│   ├── vectordb.py
│   ├── ingest.py
│   ├── pdf_loader.py
│   ├── text_splitter.py
│   ├── requirements.txt
│   └── .env.example
│
├── openai/                  # OpenAI GPT provider
│   ├── app.py
│   ├── qa_bot.py
│   ├── embeddings.py
│   ├── retriever.py
│   ├── vectordb.py
│   ├── ingest.py
│   ├── pdf_loader.py
│   ├── text_splitter.py
│   ├── requirements.txt
│   └── .env.example
│
├── google/                  # Google Gemini provider
│   ├── app.py
│   ├── qa_bot.py
│   ├── embeddings.py        # Google embeddings
│   ├── embeddings_openai.py # OpenAI embeddings (alternative)
│   ├── retriever.py
│   ├── vectordb.py
│   ├── ingest.py
│   ├── pdf_loader.py
│   ├── text_splitter.py
│   ├── requirements.txt
│   └── .env.example
│
├── hugging_face/            # HuggingFace (API + local)
│   ├── app.py
│   ├── qa_bot.py
│   ├── embeddings.py        # sentence-transformers (local)
│   ├── retriever.py
│   ├── vectordb.py
│   ├── ingest.py
│   ├── pdf_loader.py
│   ├── text_splitter.py
│   ├── requirements.txt
│   └── .env.example
│
├── watsonX/                 # IBM Watsonx provider
│   ├── app.py
│   ├── qa_bot.py
│   ├── embeddings.py        # HuggingFace (local)
│   ├── retriever.py
│   ├── vectordb.py
│   ├── ingest.py
│   ├── pdf_loader.py
│   ├── text_splitter.py
│   ├── requirements.txt
│   └── .env.example
│
├── multi/                   # Dynamic multi-provider mode
│   ├── app.py               # Streamlit UI
│   ├── qa_bot_multi.py      # Picks provider from PROVIDER env var
│   ├── embeddings_multi.py
│   ├── retriever.py
│   ├── vectordb.py
│   ├── ingest.py
│   ├── pdf_loader.py
│   ├── text_splitter.py
│   ├── requirements.txt
│   └── .env.example
│
├── fastAPI/                 # REST API + logging backend
│   ├── api.py               # FastAPI app (ask, ingest, dashboard endpoints)
│   ├── qa_bot.py
│   ├── embeddings.py
│   ├── retriever.py
│   ├── vectordb.py
│   ├── ingest.py
│   ├── pdf_loader.py
│   ├── text_splitter.py
│   ├── requirements.txt
│   └── .env.example
│
├── loggers/                 # Shared unified logging module
│   ├── __init__.py          # Exports UnifiedLogger, FileLogger, WandbLogger, MLflowLogger
│   ├── unified_logger.py    # Core logger — always writes JSONL, optionally W&B / MLflow
│   ├── log_reader.py        # Reads unified_events.jsonl for the dashboard API
│   ├── file_logger.py
│   ├── mlflow_logger.py
│   └── wandb_logger.py
│
├── dashboard/               # React + Vite monitoring dashboard
│   ├── src/
│   │   ├── components/      # StatCard, QueryLog, SessionList, ErrorList
│   │   ├── pages/           # Dashboard, Sessions, Queries, Errors
│   │   └── services/api.ts  # Fetch helpers for FastAPI logging endpoints
│   ├── package.json
│   └── vite.config.ts
│
├── chainlit.py              # Chainlit chat interface (provider set via QUOT_DIR)
├── requirements.txt         # Root-level dep for chainlit.py
├── check_deps.py            # Dependency checker — reports missing/outdated packages
├── pyrightconfig.json       # Pylance/Pyright path config for VSCode
├── .pylintrc                # Pylint config (source-path init-hook)
├── .gitignore
├── README.md
└── docs/
    ├── logging_guide.md
    ├── PROVIDER_COMPARISON.md
    └── SWITCHING_GUIDE.md

Quick Start

1. Clone the repo

git clone https://github.com/MnkyBr8n/QUOT.git
cd QUOT

2. Choose a provider and install dependencies

cd openai          # or anthropic, google, hugging_face, watsonX, multi
pip install -r requirements.txt

To verify all packages (including pip itself) are installed at the required versions:

python check_deps.py            # check all providers
python check_deps.py openai     # check one provider
python check_deps.py --install  # check and auto-install missing/outdated

3. Configure environment

cp .env.example .env
# Fill in your API keys in .env

4. Ingest a PDF

python ingest.py path/to/document.pdf

5. Run the Streamlit UI

streamlit run app.py

Configuration

OpenAI

OPENAI_API_KEY=sk-...
LLM_MODEL=gpt-4o
EMBEDDING_MODEL=text-embedding-3-small

Anthropic

ANTHROPIC_API_KEY=sk-ant-...
EMBEDDING_PROVIDER=openai          # or: google (Anthropic has no embedding API)
OPENAI_API_KEY=sk-...              # required when EMBEDDING_PROVIDER=openai
# GOOGLE_API_KEY=...               # required when EMBEDDING_PROVIDER=google
LLM_MODEL=claude-3-5-sonnet-20241022
EMBEDDING_MODEL=text-embedding-3-small

Google

GOOGLE_API_KEY=...
LLM_MODEL=gemini-2.0-flash
EMBEDDING_MODEL=models/embedding-001

HuggingFace

# API mode (recommended)
HUGGINGFACE_API_KEY=hf_...
USE_LOCAL_MODEL=false
LLM_MODEL=mistralai/Mistral-7B-Instruct-v0.2

# Local mode (no API key needed)
USE_LOCAL_MODEL=true
LLM_MODEL=TinyLlama/TinyLlama-1.1B-Chat-v1.0
DEVICE=cpu

IBM Watsonx

WATSONX_URL=https://us-south.ml.cloud.ibm.com
WATSONX_APIKEY=...
WATSONX_PROJECT_ID=...
LLM_MODEL=mistralai/mixtral-8x7b-instruct-v01
EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2

Multi-Provider

PROVIDER=openai    # openai | anthropic | google | watsonx
OPENAI_API_KEY=sk-...
LLM_MODEL=gpt-3.5-turbo
EMBEDDING_MODEL=text-embedding-3-small

# When PROVIDER=anthropic, also set:
# EMBEDDING_PROVIDER=openai   # or: google

FastAPI Service

The FastAPI module exposes a REST API and serves data for the React dashboard.

Run the server

cd fastAPI
pip install -r requirements.txt
cp .env.example .env
# Ingest a PDF first, then:
uvicorn api:app --reload

CORS

Allowed origins are controlled by ALLOWED_ORIGINS in .env (comma-separated). Default: http://localhost:5173,http://localhost:3000

Endpoints

Method	Endpoint	Description
`GET`	`/`	Health check
`POST`	`/ask`	Submit a question
`POST`	`/ingest`	Upload and index a PDF
`GET`	`/status`	System status
`GET`	`/logs/stats`	Aggregated dashboard stats
`GET`	`/logs/sessions`	All sessions with summaries
`GET`	`/logs/sessions/{id}`	Detail for a specific session
`GET`	`/logs/events`	Log events (filterable)
`GET`	`/logs/queries`	Query events
`GET`	`/logs/errors`	Error events
`GET`	`/logs/response-times`	Response time series

Example

# Ask a question
curl -X POST http://localhost:8000/ask \
  -H "Content-Type: application/json" \
  -d '{"question": "What is the main topic?"}'

# Upload a PDF
curl -X POST http://localhost:8000/ingest \
  -F "file=@document.pdf"

React Dashboard

The dashboard reads from the FastAPI logging endpoints.

cd dashboard
npm install
npm run dev        # Connects to http://localhost:8000 by default

To point at a different API, update API_BASE in dashboard/src/services/api.ts.

Chainlit Interface

# Default provider: openai
chainlit run chainlit.py

# Use a different provider
QUOT_DIR=anthropic chainlit run chainlit.py

QUOT_DIR must be the name of a provider folder (openai, anthropic, google, hugging_face, watsonX). Make sure the provider's chroma_db is already populated via ingest.py.

Unified Logging

All providers can use the shared loggers/ module. It always writes a JSONL file and optionally forwards events to W&B or MLflow.

from loggers import UnifiedLogger

logger = UnifiedLogger(
    log_dir="logs",
    vendor="openai",
    app="my-app",
    enable_wandb=False,   # set True + WANDB_API_KEY to enable
    enable_mlflow=False,  # set True to enable
)

# Wrap your QA chain
logger.log_query(query="...", answer="...", sources=[], response_time=1.2)
logger.finish()

Events are written to logs/unified_events.jsonl and consumed by both the FastAPI dashboard endpoints and the React dashboard.

Cost Reference

Provider	Input (per 1M tokens)	Output (per 1M tokens)	Context
GPT-4o	$2.50	$10.00	128K
GPT-3.5 Turbo	$0.50	$1.50	16K
Claude 3.5 Sonnet	$3.00	$15.00	200K
Claude 3 Haiku	$0.25	$1.25	200K
Gemini 2.0 Flash	$0.075	$0.30	1M
Gemini 1.5 Pro	$1.25	$5.00	2M
HuggingFace API	Free tier	Free tier	Varies
IBM Watsonx	Pay-per-token	Pay-per-token	Varies

Troubleshooting

Issue	Solution
`Missing API key`	Ensure `.env` exists in the provider folder with a real key
`Vector database not found`	Run `python ingest.py document.pdf` first
`Module not found`	Run `pip install -r requirements.txt` from the provider folder
`CORS error in dashboard`	Add your origin to `ALLOWED_ORIGINS` in `fastAPI/.env`
`chainlit: cannot import qa_bot`	Set `QUOT_DIR=<provider>` before running
Rate limit / quota exhausted	Wait or switch provider via `multi/`

Reset the vector database

rm -rf chroma_db
python ingest.py document.pdf

Contributing

Fork the repository
Create a feature branch (git checkout -b feature/new-provider)
Commit your changes
Push and open a Pull Request

License

MIT License — see LICENSE for details.

Built With

LangChain — LLM application framework
ChromaDB — Vector database
Streamlit — Web UI
FastAPI — REST API
Chainlit — Chat interface
Vite + React — Dashboard frontend

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
anthropic		anthropic
dashboard		dashboard
docs		docs
fastAPI		fastAPI
google		google
hugging_face		hugging_face
loggers		loggers
multi		multi
openai		openai
watsonX		watsonX
.gitignore		.gitignore
.markdownlint.json		.markdownlint.json
.pylintrc		.pylintrc
README.md		README.md
chainlit.py		chainlit.py
check_deps.py		check_deps.py
pyrightconfig.json		pyrightconfig.json
requirements.txt		requirements.txt

Folders and files

Latest commit

History

Repository files navigation

QUOT — Quary Unified Observability Tool

Features

Supported Providers

Project Structure

Quick Start

1. Clone the repo

2. Choose a provider and install dependencies

3. Configure environment

4. Ingest a PDF

5. Run the Streamlit UI

Configuration

OpenAI

Anthropic

Google

HuggingFace

IBM Watsonx

Multi-Provider

FastAPI Service

Run the server

CORS

Endpoints

Example

React Dashboard

Chainlit Interface

Unified Logging

Cost Reference

Troubleshooting

Reset the vector database

Contributing

License

Built With

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Uh oh!

Contributors

Uh oh!

Languages

Packages