MCP‑Stack

MCP-Stack is an MCP server orchestration project that packages ten distinct Model Context Protocol services into a unified Docker Compose environment. It connects AI development environments and local LLM clients—such as Google Gemini CLI, Claude, Cursor, and VS Code AI—to a wide range of operational tools. The included servers provide integrations for Weaviate vector memory, GitHub, local filesystem management, Supabase databases, Obsidian vaults, Zotero reference management, documentation generation, Hugo static sites, Ollama local model execution, and Fabric workflow patterns. Developers, researchers, and system architects use MCP-Stack to establish a standardized local toolchain for AI agents without configuring each MCP server individually. By grouping configurations into a centralized environment file and orchestrating deployments via Compose, it simplifies running multi-tool agent workflows locally.

Category: AI & LLM Tooling

Tags: docker, docker-compose, management, orchestration

Visit MCP‑Stack

How to install and configure MCP‑Stack

1. Clone the repository and navigate into the project directory: git clone https://github.com/neaigd/mcp_stack mcp-stack && cd mcp-stack 2. Create your environment configuration file from the template: cp .env.example .env 3. Edit .env to define your specific paths and tokens (e.g., GITHUB_PAT, PROJECTS_ROOT, VAULT_PATH). 4. Start all MCP containers in detached mode: ./scripts/start_all.sh 5. If using Google Gemini CLI, import the preconfigured endpoint configuration: gemini settings import .gemini/settings.json 6. For other MCP clients such as Claude Desktop or Cursor, configure HTTP server connections using the respective local ports defined in .gemini/settings.json (ranging from http://localhost:9100 to http://localhost:9109).

What you can do with MCP‑Stack

• Running local vector memory searches through the Weaviate MCP container to augment LLM reasoning. - Orchestrating automated documentation creation by reading project files and compiling output through docgen-mcp. - Querying and updating personal knowledge graphs directly from Claude or Gemini using the Obsidian and Zotero integrations. - Managing static website generation and automated deployments using Hugo and GitHub MCP endpoints. - Interfacing with local models via Ollama-mcp to process sensitive data without external network requests.

Key facts

• Open Source
• https://github.com/neaigd/mcp_stack
• AI & LLM Tooling, Cloud & Infrastructure, DevOps, CI/CD & Version Control
• docker, docker-compose, management, orchestration

Part of MCP Servers

What is MCP-Stack?

MCP-Stack is a centralized Docker Compose collection created by Luiz Peixoto that bundles ten Model Context Protocol servers. It provides preconfigured environments for services including vector storage, local filesystems, GitHub, Obsidian, Supabase, and Ollama under a single orchestration setup.

Which MCP clients work with MCP-Stack?

MCP-Stack works with any client supporting HTTP-based MCP endpoints. Preconfigured settings are provided for Google Gemini CLI, and the individual port mappings allow integration with Claude, Cursor, VS Code AI, and custom agent runtimes.

How do I stop all running MCP-Stack services?

You can terminate all active server containers simultaneously by executing the included stop script: ./scripts/stop_all.sh. This script invokes Docker Compose down using the centralized compose.all.yml configuration.

Is MCP-Stack open source?

Yes, MCP-Stack is open-source software distributed under the MIT license. You can inspect, modify, and self-host the orchestration scripts and Docker Compose configurations freely from the GitHub repository.