██╗   ██╗███████╗ ██████╗████████╗ ██████╗ ██████╗ ██████╗ ██████╗ ██╗██████╗ ███████╗
██║   ██║██╔════╝██╔════╝╚══██╔══╝██╔═══██╗██╔══██╗██╔══██╗██╔══██╗██║██╔══██╗██╔════╝
██║   ██║█████╗  ██║        ██║   ██║   ██║██████╔╝██║  ██║██████╔╝██║██████╔╝█████╗
╚██╗ ██╔╝██╔══╝  ██║        ██║   ██║   ██║██╔══██╗██║  ██║██╔══██╗██║██╔═══╝ ██╔══╝
 ╚████╔╝ ███████╗╚██████╗   ██║   ╚██████╔╝██║  ██║██████╔╝██████╔╝██║██║     ███████╗
  ╚═══╝  ╚══════╝ ╚═════╝   ╚═╝    ╚═════╝ ╚═╝  ╚═╝╚═════╝ ╚═════╝ ╚═╝╚═╝     ╚══════╝

Enterprise GenAI ETL Pipeline — Terminal UI

What is it?

vectordbpipe-tui is the terminal command center for the vectorDBpipe enterprise RAG SDK.

It gives you a full interactive interface — right in your terminal — to ingest documents, query with 4 AI engines, explore knowledge graphs, and manage your entire RAG pipeline without writing a single line of code.

┌────────────────────────────────────────────────────────────────────┐
│  User                                                              │
│    ↓  npm install -g vectordbpipe-tui  (auto-installs Python SDK) │
│  TUI (Node.js / Ink / React)                                       │
│    ↓  spawns subprocess                                            │
│  vdb_runner.py  →  from vectorDBpipe import VDBpipe                │
│    ↓  locally                                                      │
│  VDBpipe — 4 engines, Semantic OmniRouter, Graph, Persistence      │
└────────────────────────────────────────────────────────────────────┘

No backend server required. The TUI runs the Python SDK directly as a local subprocess.

Installation

npm install -g vectordbpipe-tui

What happens automatically:

1. ✔ Checks your Python version (requires ≥ 3.8)
2. ✔ If Python missing → shows platform-specific install guide
3. ✔ Installs vectordbpipe from PyPI automatically
4. ✔ Verifies the install
5. ✔ Prints launch instructions

Quick Start

# Launch the full TUI
vdb

# Or run specific commands
vdb setup          # Setup Wizard — configure embedder, VectorDB, LLM
vdb ingest ./data/ # Ingest documents into your pipeline
vdb query "..."    # Ask a question about your data

Features

| Feature | Description | |---|---| | Setup Wizard | Configure embedder, vector DB, and LLM step-by-step. API keys validated before saving. | | System Doctor | 6 live runtime checks: Node.js, Python, pip package, config file, internet, VectorDB config. | | Ingest | Point at a file or folder. Watch Tri-Processing run live (vectors + PageIndex + Graph). | | Chat | Semantic OmniRouter automatically picks the best engine. Badge shows which engine answered. | | Knowledge Graph | Browse entity-relationship pairs extracted from your documents. | | PageIndex | Explore the structural document index built during ingestion. | | Structured Extract | Run Engine 4 (LangChain) with a custom JSON schema. |

Engines

| Engine | Best For | Trigger | |---|---|---| | E1 — Vector RAG | Specific facts, clauses, lookups | Default fallback | | E2 — Vectorless RAG | Summaries, overviews, chapter comprehension | "summarize", "overview" | | E3 — GraphRAG | Entity relationships, multi-hop reasoning | "related", "connected", "between" | | E4 — Structured Extract | JSON output with a custom schema | vdb extract |

Architecture

vectordbpipe-tui (Node.js/Ink)
  ├── ui.js              Main React/Ink app
  ├── ui/views/          ChatView, IngestView, GraphView, etc.
  ├── lib/
  │   ├── python-bridge.js  Spawns vdb_runner.py as subprocess
  │   └── config-manager.js Reads/writes config.yaml
  └── scripts/
      ├── postinstall.cjs   Themed auto-install of Python SDK
      └── vdb_runner.py     Python CLI bridge → imports VDBpipe directly

Requirements

• Node.js ≥ 16
• Python ≥ 3.8 (auto-detected, install guide shown if missing)
• vectordbpipe PyPI package (auto-installed on npm install)

Troubleshooting

Python not found Install from python.org and check "Add to PATH".

ModuleNotFoundError: No module named 'vectorDBpipe'

python -m pip install vectordbpipe

pip launcher error on Windows (conda environments)

python -m pip install vectordbpipe

Always use python -m pip instead of pip directly on Windows.

Need help? Open an issue: https://github.com/yashdesai023/vectorDBpipe/issues

Author: Yash Desai License: MIT PyPI package: vectordbpipe