convez.ai
MCP ServerDevOps & InfraestruturaIntermediário
LikhinMN

Memspace

LikhinMN/memspace

Indexador de código ultrarrápido com MCP para assistentes de IA

Ver no GitHub
InicialPluginScore · de 0 a 100

Motor de inteligência de repositório ultrarrápido que indexa sua base de código via Tree-sitter e expõe símbolos e relacionamentos a assistentes de IA através do Model Context Protocol (MCP). Funciona como um "cérebro gráfico" local para evitar alucinações de IA e fornecer navegação precisa em milissegundos.

Instalação

$npx convez-ai LikhinMN/memspace

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
57%
Último commit
há 5 meses
Versão
v0.1.8
Issues abertas
0
Contribuidores
1

Casos de uso

  • Evitar alucinações de IA sobre nomes de funções e símbolos
  • Mapear relacionamentos e dependências de código em tempo real
  • Analisar blast radius de mudanças em funções específicas
  • Integrar contexto preciso de repositório em Claude, Cursor e outros agentes

Veredito

Solução sólida e de alto desempenho para quem trabalha com agentes de IA em grandes codebases. A implementação em C com SQLite garante velocidade, e a conformidade com MCP oferece integração nativa com ferramentas populares. Recomendado para times que precisam de precisão na análise de código.

README

A tradução para o português deste README ainda está na fila.

Memspace

Memspace is a blazing-fast, lightweight Repository Intelligence Engine built in C. It parses your codebase using Tree-sitter to extract symbols and structural relationships, and exposes them to AI coding agents via a Model Context Protocol (MCP) server.

If you are tired of AI assistants hallucinating function names or aimlessly searching your codebase, Memspace acts as a perfectly accurate, sub-millisecond local graph brain for your AI.

Features

  • 🚀 Extremely Fast: Written entirely in C with an optimized SQLite backend.
  • 🌳 True Syntax Understanding: Parses C, Python, and JavaScript ASTs natively via Tree-sitter.
  • 🔍 Relationship Mapping: Not just definitions! Automatically tracks calls across functions to determine impact radius.
  • 🤖 Native MCP Server: Fully compliant with the JSON-RPC 2.0 Model Context Protocol to plug directly into Claude, Cursor, and other AI agents.

Installation

Via NPM (Recommended)

Pre-compiled binaries are available for Windows, macOS, and Linux across x64 and ARM64 architectures.

npm install -g memspace

Building from Source

If you prefer to compile the C source yourself, ensure you have gcc and make installed:

git clone https://github.com/your-username/memspace.git
cd memspace
make clean && make

The binary will be compiled to bin/memspace.

Quick Start

  1. Initialize the workspace in the root of your target repository:
memspace init
  1. Index your codebase:
memspace index
  1. Keep it updated: As you write new code, incrementally update the index instead of rebuilding it from scratch:
memspace update

Connecting to your AI Assistant (MCP)

To give your AI assistant access to Memspace's intelligence, you need to register it as an MCP server (often called a "Skill" or "Tool"). Because we natively support JSON-RPC 2.0, it plugs directly into major AI clients.

For Claude Desktop

Add the following to your claude_desktop_config.json file:

{
  "mcpServers": {
    "memspace": {
      "command": "npx",
      "args": ["-y", "memspace", "serve"]
    }
  }
}

For Cursor IDE

  1. Open Cursor Settings -> Features -> MCP.
  2. Click + Add new MCP server.
  3. Set Type to command.
  4. Set Name to memspace.
  5. Set Command to npx -y memspace serve.
  6. Click Save.

For Roo Code (VS Code Extension)

Click the MCP Server icon in the Roo Code chat panel and add:

{
  "mcpServers": {
    "memspace": {
      "command": "npx",
      "args": ["-y", "memspace", "serve"]
    }
  }
}

Once configured, your AI will have access to powerful tools like:

  • get_symbol: Fetches exact file, line, and signature of a symbol.
  • get_callers: Finds every place a function is called.
  • impact_of: Analyzes the direct and transitive blast radius of changing a specific function.
  • find_feature: Performs a fuzzy text search across all symbols.

Development

See CONTRIBUTING.md for details on how to build, test, and contribute to Memspace.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Ver o original no GitHub

Bom para

Relacionados