Minima
On-premises conversational RAG with configurable containers
What is Minima?
Minima is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to on-premises conversational rag with configurable containers
On-premises conversational RAG with configurable containers
This server falls under the Data Science & ML category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- On-premises conversational RAG with configurable containers
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx minimaConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Minima
Minima is an on-premises conversational RAG (Retrieval-Augmented Generation) system delivered as Docker containers that index local files — PDFs, Word documents, Excel sheets, Markdown, CSV, and plain text — and expose a chat interface for querying them with natural language. It supports fully local operation via Ollama, custom OpenAI-compatible LLM endpoints, and Anthropic Claude via MCP integration. Privacy-conscious organizations use Minima to build internal knowledge bases that keep sensitive documents off external cloud services while still benefiting from LLM-powered question answering.
Prerequisites
- Docker and Docker Compose installed
- Python 3.10 or later and `uv` package manager (for MCP mode: `pip install uv`)
- A folder of documents to index (supports .pdf, .docx, .xls, .txt, .md, .csv)
- For Ollama mode: Ollama installed and a model pulled locally
- For ChatGPT/Claude mode: an OpenAI API key or Anthropic Claude MCP client
Clone the repository
Clone the Minima repository and enter the project directory.
git clone https://github.com/dmayboroda/minima.git
cd minimaCreate your environment file
Copy the example `.env` file and configure the required variables. Set `LOCAL_FILES_PATH` to the absolute path of your document folder, choose an `EMBEDDING_MODEL_ID`, and set the appropriate model variables for your chosen deployment mode.
cp .env.example .env
# Edit .env:
# LOCAL_FILES_PATH=/home/user/documents
# EMBEDDING_MODEL_ID=sentence-transformers/all-MiniLM-L6-v2
# EMBEDDING_SIZE=384Choose a deployment mode and start
Pick the Docker Compose file matching your setup. Use `docker-compose-ollama.yml` for a fully local stack, `docker-compose-custom-llm.yml` for any OpenAI-compatible endpoint, or `docker-compose-mcp.yml` for Claude Desktop integration.
# Fully local with Ollama
docker compose -f docker-compose-ollama.yml --env-file .env up --build
# Custom LLM (vLLM, LiteLLM, etc.)
docker compose -f docker-compose-custom-llm.yml --env-file .env up --build
# MCP mode for Claude Desktop
docker compose -f docker-compose-mcp.yml --env-file .env up --buildWait for indexing to complete
On first startup Minima indexes all files in `LOCAL_FILES_PATH`. Watch the container logs — indexing can take several minutes for large document sets.
docker compose logs -fOpen the chat interface
Navigate to `http://localhost:3000` in your browser to start querying your documents through the conversational UI.
Connect to Claude Desktop via MCP (optional)
If you started with `docker-compose-mcp.yml`, add the MCP server endpoint to your Claude Desktop config and restart the client to enable document Q&A from within Claude.
Minima Examples
Client configuration
Claude Desktop MCP configuration when running Minima's MCP Docker stack locally on port 8080.
{
"mcpServers": {
"minima": {
"command": "uvx",
"args": ["minima-mcp"],
"env": {
"MINIMA_URL": "http://localhost:8080"
}
}
}
}Prompts to try
Sample prompts once Minima is running and documents are indexed.
- "What does our employee handbook say about remote work policies?"
- "Summarize the key financial figures from the Q3 2024 report."
- "Find all mentions of 'data retention' in my indexed documents."
- "Compare the specifications in the two product datasheets I uploaded."
- "What are the action items from last month's meeting notes?"Troubleshooting Minima
Containers start but the web UI shows no results for queries
Check that `LOCAL_FILES_PATH` in `.env` is an absolute path that exists and contains supported files. View indexing progress with `docker compose logs minima-indexer`. If the path is a relative path or doesn't exist, the indexer silently processes zero documents.
Ollama mode fails with model not found errors
Ensure Ollama is running and the model specified in `OLLAMA_MODEL` has been pulled: `ollama pull llama3`. The Docker Compose Ollama stack expects Ollama to be accessible at `http://host.docker.internal:11434` — on Linux, set `extra_hosts: ["host.docker.internal:host-gateway"]` in the compose file.
Large PDF files are not indexed or cause memory errors
Increase Docker's memory limit to at least 4 GB in Docker Desktop settings. Very large PDFs (100+ pages) may need to be split before indexing. Check the `EMBEDDING_SIZE` matches the chosen `EMBEDDING_MODEL_ID` — a mismatch causes silent failures during vector storage.
Frequently Asked Questions about Minima
What is Minima?
Minima is a Model Context Protocol (MCP) server that on-premises conversational rag with configurable containers It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Minima?
Follow the installation instructions on the Minima GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Minima?
Minima works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Minima free to use?
Yes, Minima is open source and available under the MPL-2.0 license. You can use it freely in both personal and commercial projects.
Minima Alternatives — Similar Data Science & ML Servers
Looking for alternatives to Minima? Here are other popular data science & ml servers you can use with Claude, Cursor, and VS Code.
Ultrarag
★ 5.6kA Low-Code MCP Framework for Building Complex and Innovative RAG Pipelines
RocketRide
★ 3.1k📇 🏠 - MCP server that exposes RocketRide AI pipelines as t
Aix Db
★ 2.1kAix-DB 基于 LangChain/LangGraph 框架，结合 MCP Skills 多智能体协作架构，实现自然语言到数据洞察的端到端转换。
NeMo Data Designer
★ 1.9k🎨 NeMo Data Designer: Generate high-quality synthetic data from scratch or from seed data.
PaperBanana
★ 1.7kOpen source implementation and extension of Google Research’s PaperBanana for automated academic figures, diagrams, and research visuals, expanded to new domains like slide generation.
MiniMax
★ 1.5kBridges MiniMax AI capabilities to the Model Context Protocol, enabling AI agents to perform image understanding, text-to-image generation, and speech synthesis. It provides a standardized interface for accessing MiniMax's core tools via JSON-RPC.
Browse More Data Science & ML MCP Servers
Explore all data science & ml servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Minima in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Minima?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.