https://github.com/spirita1204/mcp-client
implementation for interacting between an MCP client and an Ollama model
https://github.com/spirita1204/mcp-client
llm mcp ollama python
Last synced: 3 months ago
JSON representation
implementation for interacting between an MCP client and an Ollama model
- Host: GitHub
- URL: https://github.com/spirita1204/mcp-client
- Owner: spirita1204
- License: mit
- Created: 2025-03-24T16:17:16.000Z (3 months ago)
- Default Branch: master
- Last Pushed: 2025-03-24T17:03:09.000Z (3 months ago)
- Last Synced: 2025-03-24T17:36:09.741Z (3 months ago)
- Topics: llm, mcp, ollama, python
- Language: Python
- Homepage:
- Size: 26.4 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# MCPClient Python Application
This is a Python client application designed to interact with an MCP (Model Context Protocol) server.
## Features
- **Asynchronous Communication:** Uses `asyncio` for non-blocking communication between the client and server.
- **Customizable Server Scripts:** The client can connect to both Python and JavaScript-based server scripts.
- **Tool Management:** Dynamically fetches and interacts with tools available on the connected server.
- **Chat Interface:** Provides a simple command-line interface to interact with the server in a conversational format.
- **Tool Integration:** Supports extracting JSON-formatted tool calls from server responses and executing them.
- **Environment Variable Loading:** Supports loading environment variables from a `.env` file using the `dotenv` package.
## Requirements
- Python 3.7 or higher
- `asyncio` library (included with Python)
- `requests` for HTTP requests to the server
- `mcp` (custom library for handling MCP communication)
- `dotenv` for environment variable management
## Setup
1. **Clone the repository** (or download the script files) to your local machine.
2. **Install required dependencies:**
```bash
pip install -r requirements.txt
1. **Create a `.env` file** in the root directory to load necessary environment variables. For example:
```
BASE_URL=http://localhost:11434
MODEL=llama3.2
```
2. **Run the client** with the path to the server script:
```bash
python client.py
```
The server script can be a Python `.py` or JavaScript `.js` file.
## How It Works
1. **Connecting to the MCP Server:** The client connects to the server via standard input/output channels, using the provided script (`.py` or `.js`).
2. **Processing Queries:** The client sends user queries to the server and receives responses. Available tools are listed and can be called directly from the assistant’s replies.
3. **Tool Execution:** If a response contains a valid tool call (in JSON format), the client extracts the call and triggers the respective tool on the server.
4. **Interaction:** The client interacts with the server in a conversational format, displaying results from server tools and continuing the conversation.
## Example Workflow
1. The user enters a query like:
```bash
Question: What is the weather today?
```
2. The client sends the query to the server, which responds with available tools and information.
3. If the server suggests using a weather tool, the client executes the tool with the necessary parameters and shows the result.
4. The client continues the conversation based on the new information returned by the tool.