vectra

2 min read

Vectra is a lightweight terminal-based interface for managing vector stores - starting with ChromaDB, with plans to support other backends. It helps you create, query, and organize collections of embedded documents through an intuitive TUI.

Features

• Query documents via semantic similarity
• Create, list, and delete collections
• View stored documents
• Powered by ChromaDB with pluggable vector DB support (soon)
• Easy to integrate into other Python tools (e.g., web frameworks)

Getting Started

Installation

git clone https://github.com/yourusername/vectra.git
cd vectra

# using pip
pip install -r requirements.txt

# using uv
uv add -r requirements.txt
````

> Make sure you have Python 3.8+ and `chromadb` installed.

### Running

```bash
python main.py

# or
uv run main.py

You’ll be greeted with an interactive interface where you can create collections, add/query documents, and manage your vector DB.

Tech Stack

• Python 3.8+
• ChromaDB
• Rich for terminal rendering
• Questionary for interactive prompts

Roadmap

Here’s what we plan to build next:

• [x] Basic TUI structure with Rich
• [x] Integration with ChromaDB
• [x] Collection creation/deletion/listing
• [x] Basic querying via CLI
• [ ] Add document support (with metadata)
• [ ] Multi-result querying
• [ ] Pluggable embedder support (sentence-transformers, OpenAI, Cohere, etc.)
• [ ] Multiple vector DB backends (Qdrant, Weaviate)
• [ ] Export/import collections
• [ ] Search history
• [ ] Optional CLI (non-TUI) interface
• [ ] Optional web mode via FastAPI wrapper
• [ ] PyPI release (pip install vectra)

Contributing

Pull requests, ideas, and feedback are welcome! Just open an issue or fork the repo.