Weaver

Python project scaffolding tool that generates entity-resolution projects from bundled cookiecutter templates.

Features

• Generate complete entity-resolution projects with a single command
• Multiple project templates optimized for different use cases
• LLM-powered ontology generation for knowledge graphs
• Integrated support for databases, orchestrators, APIs, and search engines
• Docker and testing support out of the box

Installation

pip install weaver

Or install from source:

git clone https://github.com/yourusername/weaver.git
cd weaver
uv sync

Quick Start

Create a new project:

# Create a knowledge graph project
weaver create --template knowledge-graph --name my-project

# Create an advanced search project
weaver create --template advanced-search --name my-search-app

# List available templates
weaver list-templates

CLI Commands

create

Generate a new project from a template.

weaver create --template <template-name> --name <project-name> [options]

Options:

• --name, -n - Name of your project
• --template, -t - Project template to use (see available templates below)
• --list-deps - Show dependencies without creating project

Knowledge Graph Options:

• --ontology / --no-ontology - Enable/disable ontology generation (default: enabled)
• --llm anthropic|openai - LLM provider for ontology generation (default: anthropic)
• --db-host - Database host for schema inspection (default: localhost)
• --db-port - Database port for schema inspection (default: 5432)

Examples:

# Basic knowledge graph project
weaver create --template knowledge-graph --name company-kg

# Knowledge graph with OpenAI for ontology generation
weaver create --template knowledge-graph --name my-kg --llm openai

# Advanced search project
weaver create --template advanced-search --name search-engine

# Show dependencies without creating project
weaver create --template knowledge-graph --name test --list-deps

list-templates

List all available project templates.

weaver list-templates

templates

Show detailed information about available templates.

weaver templates

version

Show Weaver version.

weaver version

Available Templates

knowledge-graph

Build knowledge graphs with Neo4j, web scraping, and LLM-powered ontology generation.

Includes:

• Neo4j graph database
• Web scraping and API data extraction
• Prefect orchestration
• FastAPI endpoints
• NLP processing
• LLM-powered ontology designer
• Schema inspection tools

Best for: Entity resolution, knowledge graphs, relationship mapping

See the full demo: Knowledge Graph Demo Guide

advanced-search

Hybrid sparse and vector search system optimized for complex queries.

Includes:

• PostgreSQL with vector extensions
• Hybrid vector + keyword search
• Prefect orchestration
• FastAPI search endpoints
• Web scraping

Best for: Search engines, recommendation systems, semantic search

news-analyzer

News aggregation and analysis with bias detection.

Includes:

• PostgreSQL database
• NLP processing
• Vector search
• Elasticsearch
• API and web scraping

Best for: Content analysis, news aggregation, sentiment analysis

basic

Simple entity-relationship project for quick prototyping.

Includes:

• SQLite database
• Simple orchestrator
• Basic project structure

Best for: Learning, prototyping, simple data projects

Project Structure

Generated projects follow a consistent architecture:

my-project/
├── src/
│   ├── extractors/         # Data gathering (web, API, files, DB)
│   ├── storage/            # Database integration
│   ├── pipeline/           # Orchestration workflows
│   ├── api/                # REST/search endpoints (optional)
│   ├── ontology/           # Schema inspection & ontology generation (optional)
│   └── config/
│       └── settings.py     # Configuration with env variables
├── tests/
├── pyproject.toml
├── README.md
├── .env.example
└── docker-compose.yml      # If Docker enabled

Knowledge Graph with Ontology Generation

Weaver's knowledge graph template includes powerful LLM-based ontology generation:

# Create knowledge graph project
weaver create --template knowledge-graph --name company-kg --ontology

cd company-kg

# Set your API key
export ANTHROPIC_API_KEY=sk-ant-your-key-here

# Inspect database schema
python -m src.ontology.schema_inspector 'postgresql://user:pass@localhost/mydb'

# Generate ontology from schema using Claude
python -m src.ontology.llm_designer 'postgresql://user:pass@localhost/mydb' MyOntology

The ontology generator will:

• Analyze your database schema
• Identify entity classes and relationships
• Detect natural keys for entity resolution
• Generate properties and cardinality constraints
• Provide design explanations and improvement suggestions
• Export as JSON for implementation in Neo4j

Complete walkthrough: See DEMO_KNOWLEDGE_GRAPH.md for a full step-by-step demo with sample data.

Configuration

Each template supports different configurations:

• Databases: SQLite, PostgreSQL, Neo4j, MongoDB
• Orchestrators: Prefect, Dagster, Simple
• Search Engines: Vector Hybrid, Elasticsearch, None
• API Frameworks: FastAPI, Flask, Django, None
• Features: Vector search, NLP, web scraping, API scraping, Docker, pytest

Development

# Install dependencies
uv sync

# Run tests
pytest

# Run CLI locally
uv run weaver --help

# Build package
uv build

Requirements

• Python 3.8+
• PostgreSQL (for ontology generation and some templates)
• Neo4j (for knowledge graph template)
• LLM API key (Anthropic or OpenAI) for ontology generation

License

MIT

Contributing

Contributions welcome! Please open an issue or submit a pull request.

Resources

• Knowledge Graph Demo - Complete walkthrough with ontology generation
• Cookiecutter Documentation
• Neo4j Graph Database
• Prefect Workflows
• Anthropic API