Weaviate-Spy

Weaviate-Spy is a web client for exploring local docker Weaviate Weaviate.io vector database.

Features

• Health Check - Monitor Weaviate connection status
• Schema Browser - View all collections and their properties
• Collections Browser - Browse and search collection data
  • Aggregations - Group and count by property
  • Generative Search - RAG with Ollama LLMs
    • Semantic Search - AI-powered near-text with certainty
    • Keyword Search - BM25 full-text search
    • Hybrid Search - Combine BM25 + vector with alpha

Quick Start

# Clone the repository
git clone https://github.com/AISidesKicks/weaviate-spy.git
cd weaviate-spy

# Start the application - build container
docker compose up -d --build

Open http://localhost:7777 in your browser.

Usage

See compose.yml and adjust environment variables as needed. By default it connects to a locally hosted (docker) Weaviate on port 8080 (without auth credentials).

Dummy Data / Testing

See dummy/dummy.md for setting up test data with Weaviate and Ollama.

The dummy directory contains:

• docker-compose.yml - Weaviate + Ollama setup on Nvidia (RTX 4060 12GB VRAM)
• ollama_test.sh - Script to test Ollama models (embeding and generation)
• weaviate_dummy.py - Script to create sample Filmy collection and to test it

More screenshots

Inspiration

Original simple Web UI for Weaviate V3: naaive/weaviate-ui

Disclaimer

This project is for educational purposes only and is not affiliated with or sponsored by Weaviate or Ollama. Use at your own risk.

CAUTION: There is no authorization or authentication in place in Weaviate Spy! Use only in trusted local environments.

Contribution

Any form of contribution is welcome, including but not limited to submitting bug reports, proposing new features, improving code, etc.