Home

VaultMind Wiki

Overview

VaultMind is a production-oriented Retrieval-Augmented Generation (RAG) system built using LangGraph, LangChain, OpenAI, ChromaDB, and BM25.

The project is designed to answer questions about a knowledge source while maintaining high retrieval quality, strong safety controls, evaluation capabilities, and production-grade engineering practices.

Core Features

• Hybrid Retrieval (ChromaDB + BM25)

• LangGraph Workflow Orchestration

• Input Guardrails

• Output Guardrails

• Query Classification

• Context Validation

• Retry Logic

• Confidence Scoring

• Token Usage Tracking

• Cost Estimation

• RAGAS Evaluation

• GitHub Actions CI

Tech Stack

Component | Technology -- | -- Workflow Engine | LangGraph LLM Framework | LangChain LLM | GPT-4o-mini Embeddings | text-embedding-3-small Vector Database | ChromaDB Keyword Search | BM25 Evaluation | RAGAS Configuration | pydantic-settings CI/CD | GitHub Actions

Milestone

Current Release:

VaultMind v1.0 — Production Ready RAG System

Target Completion Date:

19 June 2026