synapstor 0.1.4

An unofficial evolution of mcp-server-qdrant - Client and server for semantic storage and search

Synapstor 📚🔍

Version 0.1.4 | Python 3.10+ | MIT License

🌎 Idioma / Language

• Português 🇧🇷
• English 🇺🇸

---

Português 🇧🇷

Synapstor é uma biblioteca modular para armazenamento e recuperação semântica de informações usando embeddings vetoriais e banco de dados Qdrant.

Nota: O Synapstor é uma evolução não oficial do projeto mcp-server-qdrant, expandindo suas funcionalidades para criar uma solução mais abrangente para armazenamento e recuperação semântica.

🔭 Visão Geral

Synapstor é uma solução completa para armazenamento e recuperação de informações baseada em embeddings vetoriais. Combinando a potência do Qdrant (banco de dados vetorial) com modelos modernos de embeddings, o Synapstor permite:

• 🔍 Busca semântica em documentos, código e outros conteúdos textuais
• 🧠 Armazenamento eficiente de informações com metadados associados
• 🔄 Integração com LLMs através do Protocolo MCP (Model Control Protocol)
• 🛠️ Ferramentas CLI para indexação e consulta de dados

🖥️ Requisitos

• Python: 3.10 ou superior
• Qdrant: Banco de dados vetorial para armazenamento e busca de embeddings
• Modelos de Embedding: Por padrão, usa modelos da biblioteca FastEmbed

📦 Instalação

# Instalação básica via pip
pip install synapstor

# Com suporte a embeddings rápidos (recomendado)
pip install "synapstor[fastembed]"

# Para desenvolvimento (formatadores, linters)
pip install "synapstor[dev]"

# Para testes
pip install "synapstor[test]"

# Instalação completa (todos os recursos e ferramentas)
pip install "synapstor[all]"

🚀 Uso Rápido

Configuração

Existem várias formas de configurar o Synapstor:

1. Variáveis de ambiente:

   # Exportar as variáveis no shell (Linux/macOS)
   export QDRANT_URL="http://localhost:6333"
   export QDRANT_API_KEY="sua-chave-api"
   export COLLECTION_NAME="synapstor"
   export EMBEDDING_MODEL="sentence-transformers/all-MiniLM-L6-v2"
   
   # Ou no Windows (PowerShell)
   $env:QDRANT_URL = "http://localhost:6333"
   $env:QDRANT_API_KEY = "sua-chave-api"
   $env:COLLECTION_NAME = "synapstor"
   $env:EMBEDDING_MODEL = "sentence-transformers/all-MiniLM-L6-v2"
   

2. Parâmetros na linha de comando:

   synapstor-ctl start --qdrant-url http://localhost:6333 --qdrant-api-key sua-chave-api --collection-name synapstor --embedding-model "sentence-transformers/all-MiniLM-L6-v2"
   

3. Programaticamente (para uso como biblioteca):

   from synapstor.settings import Settings
   
   settings = Settings(
       qdrant_url="http://localhost:6333",
       qdrant_api_key="sua-chave-api",
       collection_name="minha_colecao",
       embedding_model="sentence-transformers/all-MiniLM-L6-v2"
   )
   

Como servidor MCP

# Iniciar o servidor MCP com a interface centralizada
synapstor-ctl start

# Com parâmetros de configuração
synapstor-ctl start --qdrant-url http://localhost:6333 --qdrant-api-key sua-chave-api --collection-name minha_colecao --embedding-model "sentence-transformers/all-MiniLM-L6-v2"

Indexação de projetos

# Indexar um projeto
synapstor-ctl indexer --project meu-projeto --path /caminho/do/projeto

Como biblioteca em aplicações Python

from synapstor.qdrant import QdrantConnector, Entry
from synapstor.embeddings.factory import create_embedding_provider
from synapstor.settings import EmbeddingProviderSettings

# Inicializar componentes
settings = EmbeddingProviderSettings()
embedding_provider = create_embedding_provider(settings)

connector = QdrantConnector(
    qdrant_url="http://localhost:6333",
    collection_name="minha_colecao",
    embedding_provider=embedding_provider
)

# Armazenar informações
async def store_data():
    entry = Entry(
        content="Conteúdo a ser armazenado",
        metadata={"chave": "valor"}
    )
    await connector.store(entry)

# Buscar informações
async def search_data():
    results = await connector.search("consulta em linguagem natural")
    for result in results:
        print(result.content)

📚 Documentação Completa

Para documentação detalhada, exemplos avançados, integração com diferentes LLMs, deployment com Docker, e outras informações, visite o repositório no GitHub.

---

English 🇺🇸

Synapstor is a modular library for semantic storage and retrieval of information using vector embeddings and the Qdrant database.

Note: Synapstor is an unofficial evolution of the mcp-server-qdrant project, expanding its functionality to create a more comprehensive solution for semantic storage and retrieval.

🔭 Overview

Synapstor is a complete solution for storing and retrieving information based on vector embeddings. Combining the power of Qdrant (vector database) with modern embedding models, Synapstor allows:

• 🔍 Semantic search in documents, code, and other textual content
• 🧠 Efficient storage of information with associated metadata
• 🔄 Integration with LLMs through the MCP (Model Control Protocol)
• 🛠️ CLI tools for indexing and querying data

🖥️ Requirements

• Python: 3.10 or higher
• Qdrant: Vector database for storing and searching embeddings
• Embedding Models: By default, uses models from the FastEmbed library

📦 Installation

# Basic installation via pip
pip install synapstor

# With fast embedding support (recommended)
pip install "synapstor[fastembed]"

# For development (formatters, linters)
pip install "synapstor[dev]"

# For testing
pip install "synapstor[test]"

# Complete installation (all features and tools)
pip install "synapstor[all]"

🚀 Quick Usage

Configuration

There are several ways to configure Synapstor:

1. Environment variables:

   # Export variables in shell (Linux/macOS)
   export QDRANT_URL="http://localhost:6333"
   export QDRANT_API_KEY="your-api-key"
   export COLLECTION_NAME="synapstor"
   export EMBEDDING_MODEL="sentence-transformers/all-MiniLM-L6-v2"
   
   # Or on Windows (PowerShell)
   $env:QDRANT_URL = "http://localhost:6333"
   $env:QDRANT_API_KEY = "your-api-key"
   $env:COLLECTION_NAME = "synapstor"
   $env:EMBEDDING_MODEL = "sentence-transformers/all-MiniLM-L6-v2"
   

2. Command line parameters:

   synapstor-ctl start --qdrant-url http://localhost:6333 --qdrant-api-key your-api-key --collection-name synapstor --embedding-model "sentence-transformers/all-MiniLM-L6-v2"
   

3. Programmatically (for use as a library):

   from synapstor.settings import Settings
   
   settings = Settings(
       qdrant_url="http://localhost:6333",
       qdrant_api_key="your-api-key",
       collection_name="my_collection",
       embedding_model="sentence-transformers/all-MiniLM-L6-v2"
   )
   

As an MCP server

# Start the MCP server with the centralized interface
synapstor-ctl start

# With configuration parameters
synapstor-ctl start --qdrant-url http://localhost:6333 --qdrant-api-key your-api-key --collection-name my_collection --embedding-model "sentence-transformers/all-MiniLM-L6-v2"

Project indexing

# Index a project
synapstor-ctl indexer --project my-project --path /path/to/project

As a library in Python applications

from synapstor.qdrant import QdrantConnector, Entry
from synapstor.embeddings.factory import create_embedding_provider
from synapstor.settings import EmbeddingProviderSettings

# Initialize components
settings = EmbeddingProviderSettings()
embedding_provider = create_embedding_provider(settings)

connector = QdrantConnector(
    qdrant_url="http://localhost:6333",
    collection_name="my_collection",
    embedding_provider=embedding_provider
)

# Store information
async def store_data():
    entry = Entry(
        content="Content to be stored",
        metadata={"key": "value"}
    )
    await connector.store(entry)

# Search for information
async def search_data():
    results = await connector.search("natural language query")
    for result in results:
        print(result.content)

📚 Complete Documentation

For detailed documentation, advanced examples, integration with different LLMs, Docker deployment, and other information, visit the GitHub repository.

---

Developed with ❤️ by the Synapstor team