ZIN MCP Client

First of all thanks to everyone who supported this project. But I have to make it Read-Only and drop future developments due to below mentioned reasons:

- Other Projects: I have to work on the other projects which are much more usefull
- Resource Limitation: Due to limited time and resources, I am not able to add more features in it
- Better Alternatives: Even if I work on this project and add many more features, the end result will not make sesnse as such features related to MCP client and all is already implemented in other alternatives, So I want to spend that time into other projects such Jadx AI MCP
- Goal Achieved: The initial goal for this project is to make cli based lightweight MCP Client so I can test my MCP Servers. Adding more features will just make it something it is not intended to become. For example, the web ui update that I implemented but they are out of place as there already exist much better alternatives.

ZIN-MCP-CLIENT (Part of Zin's MCP Suite)

⚡ Lightweight, Fast, Simple, Cross Platform, CLI, Web UI and Open Web UI based MCP Client for STDIO MCP Servers, to fill the gap and provide bridge between your local LLMs running Ollama and MCP Servers.

A powerful yet lightweight, fast, simple and flexible client for interacting with MCP (Model Context Protocol) servers through local LLMs. This tool allows you to connect to and utilize various tools from multiple MCP servers through an easy-to-use command-line interface.

https://github.com/user-attachments/assets/fad3d994-8113-47df-b10c-54541a5c3aec

https://github.com/user-attachments/assets/4cd26715-b5e6-4b4b-95e4-054de6789f42

https://github.com/user-attachments/assets/94e1ff33-0c88-40a8-8447-d9c3278a1d50

- Supports Multiple Servers -GhidraMCP Server

https://github.com/user-attachments/assets/759f7138-1cc5-400f-b7e2-a6a7f0dca894

https://github.com/user-attachments/assets/704c214b-0ebb-4da9-971c-5f04446e9646

- Multi-Server Support:Connect to multiple MCP servers simultaneously
- Local LLM Integration:Use local LLMs via Ollama for privacy and control
- Interactive CLI:Clean, intuitive command-line interface with rich formatting
- Minimal Light Weight Web UI:Clean, minimal, lightweight web ui for ease of use
- Open Web UI Integration:Plug n Play integration withOpen Web UIto get rich features ofOpen Web UIwith strong MCP Client
- Comprehensive Logging:Detailed logs for debugging and troubleshooting
- ReAct Agent Framework:Utilizes LangChain's ReAct agent pattern to intelligently invoke tools
- Cross Platform:Cross Platform Support
- Simple, Fast, Lightweight:It is simple, it is fast, it is lightweight

1. Downlaod from Releases:https://github.com/zinja-coder/zin-mcp-client/releases

# 1. unzip zin-mcp-client-<version>.zip ├zin-mcp-client/ ├── zin_mcp_client.py ├── web_client.py ├── mcp_proxy.py ├── static/ ├── src/ ├── mcp-config.json ├── README.md ├── LICENSE # 2. Navigate to zin-mcp-client directory cd zin-mcp-client # 3. This project uses uv (recommended) - https://github.com/astral-sh/uv instead of pip for dependency management. ## a. Install uv (if you dont have it yet) - (Only Required Step) curl -LsSf https://astral.sh/uv/install.sh | sh # All below steps are not required. ## b. OPTIONAL, if for any reasons, you get dependecy errors in jadx-mcp-server, Set up the environment uv venv source .venv/bin/activate # or .venv\Scripts\activate on Windows ## c. OPTIONAL Install dependencies uv pip install -r requirements.txt # 4. Not recommended, you can also use pip for this. pip install -r requirements.txt or pip install -r requirements.txt --break-system-packages # The setup for zin-mcp-client is done.

1. Download and Install ollama: https://ollama.com/download If you are on linux you can directly run below command to install it: > curl -fsSL https://ollama.com/install.sh | sh 2. Download and run any LLM that has capability to invoke tool. For example, the llama 3.1 has capability to invoke the tool. You can run it using following command: > ollama run llama3.1:8b [Note]: Kindly note the above command will fetch the model with 4b parameters. If you have stronger hardware kindly fetch higher parameter model for better performance. 3. Serve the Ollama on API server using following command > ollama serve This will serve the ollama api on port 1134, you can confirm that it running using curl command as following: > curl http://localhost:11434/ 18:54:00 Ollama is running`

{ "mcpServers": { "jadx-mcp-server": { "command": "/path/to/uv", "args": [ "--directory", "/path/to/jadx-mcp-server/", "run", "jadx_mcp_server.py" ] }, "apktool-mcp-server": { "command": "/path/to/uv", "args": [ "--directory", "/path/to/apktool-mcp-server/", "run", "apktool_mcp_server.py" ] } } }

- path/to/uvwith the actual path to youruvexecutable
- path/to/jadx-mcp-serverwith the absolute path to where you have stored the jadx-mcp-server

The default location of config file is inside zin-mcp-client directory namedmcp-config.json, however you can provide path to your own config file using--configoption such as

uv run zin_mcp_client.py --server jadx-mcp-server --model llama3.1:8b --config /home/zinjacoder/mcp-config.json

- Use--serveroption to specify server of your choice, use--configoption to provide path to your config file, use--modeloption to use specific model, use--debugto enable verbose output

- You can also use the--debug`flag to enable debug to print each and every detail on console on runtime to help you find the issue.

https://github.com/user-attachments/assets/ee478917-c4f5-46fb-9f0e-ad31d7c33ee0

- If you can not resolve the error on you own, use the logs and debug mode's output and provide them to us by opening an issue athttps://github.com/zinja-coder/zin-mcp-client/issues
- Done configuration mentioned above
- Run web client using following:
- Demo of the web client is shown in video at[top
- Done configuration mentioned above.
- Run the MCP Proxy using following:
- Go to Open Web Ui portal.
- In Open Web UI -> Settings -> Connection -> add new connection as shown in below image

Add the URL for MCP Proxy running on port 8000 with localhost if running locally as shown in image. You can put anything in API key.
- Now prompt and utilize the MCP client, Plug n Play setup.

Currently, proprietary API-based models like Anthropic’s Claude tend to be more proficient proficient at tool calling.

…