🤖 mcp-ollama-beeai

What is 🤖 mcp-ollama-beeai?

A minimal client app that interacts with local Ollama models using multiple MCP (Model Context Protocol) agent tools, built with the BeeAI framework. It provides a chat interface where you can select MCP agents – such as Postgres or Fetch – and let the LLM reason and act on your requests.

How to use 🤖 mcp-ollama-beeai?

Install Ollama, serve it locally, and pull at least one model (default: llama3.1). Configure your MCP servers in mcp-servers.json (defaults include Postgres and Fetch). Clone the repo, run npm i && npm start, then open http://localhost:3000. Use the UI dropdowns to choose a server and tool before asking questions.

Key features of 🤖 mcp-ollama-beeai

• Chat interface with LLM‑powered reasoning and action steps
• Use any local Ollama model (configurable via .env)
• Pre‑configured Postgres and Fetch MCP servers
• Switch MCP agents on the fly from the UI
• Responses rendered in readable Markdown format
• Built on the BeeAI framework for easy ReAct agent setup

Use cases of 🤖 mcp-ollama-beeai

• Query a local Postgres database using natural language
• Fetch and summarise web pages through a chat interface
• Experiment with different Ollama models paired with MCP tools
• Quickly test and demo MCP‑based agent workflows locally

FAQ from 🤖 mcp-ollama-beeai

Which Ollama models are supported?

Any model you have pulled locally (default is llama3.1). Override OLLAMA_CHAT_MODEL in .env to use a different one.

What MCP servers are included by default?

Postgres and Fetch are pre‑configured. You can add any MCP server from the official examples by editing mcp-servers.json.

What are the system requirements?

At least 16GB of RAM for models to perform well. Ollama must be installed and running, and Node.js/npm are needed to run the app.

How do I point the app to a different LLM server?

Set OLLAMA_BASE_URL in .env (default: http://localhost:11434/api) to use a remote server instead of your local Ollama.

Does the app require an internet connection?

No. All interactions are local – the Ollama model runs on your machine and the MCP agents connect to local or remote services you configure.