mcp_sample

# Google Search AI Agent This project is a sample code for an AI agent using the Google Search API. It is executed using the Python package manager, uv. ## Project Structure ### Host (AI Agent) 1. **agent_chat_with_google_search.py** * A CLI-based chat that answers based on the mcp server's fetch and Google search. ### MCP Server 1. **server_google_search.py** * A Google search server using the official Google API. It returns the top 5 Google search results. ## Prerequisites 1. Python 3.12 or higher 2. uv package manager 3. Google Search API key 4. OpenAI API key ## Environment Setup Steps ### 1. Install uv uv is a Python package manager that makes it easy to manage dependencies. ```bash # Install uv using curl curl -LsSf https://astral.sh/uv/install.sh | sh ``` ### 2. Configure Google API 1. Go to [Google Cloud Console](https://console.cloud.google.com) and create or select a project. 2. Enable the "Custom Search API" from "API & Services" → "Library". 3. Create an API key by going to "API & Services" → "Credentials" → "Create Credentials" → "API Key" (this will be your `GOOGLE_CSE_API_KEY`). 4. Visit [Programmable Search Engine](https://programmablesearchengine.google.com/about/) and create a search engine. 5. Copy the search engine ID (cx value) from the search engine settings page (this will be your `GOOGLE_CSE_ID`). *Note: The Google API is free for up to 100 queries per day; beyond that, it costs $5 per 1,000 requests (up to a maximum of 10,000 requests per day).* ### 3. Obtain OpenAI API Key 1. Create an account on [OpenAI's website](https://platform.openai.com/). 2. Issue an API key (note that this is a paid service). ### 4. Set Environment Variables Create a `.env` file in the root directory of the project, referencing the `.env.sample` file, and set the following content: ```env OPENAI_API_KEY=your_OpenAI_API_key GOOGLE_CSE_API_KEY=your_Google_API_key GOOGLE_CSE_ID=your_search_engine_ID ``` ## MCP Server Configuration In the agent file, the MCP server configuration is defined as follows: ```python RAW_CONFIG: Dict[str, dict] = { "fetch": {"command": "uvx", "args": ["mcp-server-fetch"]}, "google_search": { "command": "uv", "args": ["--directory", "/path/to/your/project/servers/src", "run", "server_google_search.py"], }, } ``` **Note**: When using it, you need to change the `/path/to/your/project/servers/src` part to the path that matches your environment. ## How to Run Execute the following command in the root directory of the project: ### Run Google Search Chatbot ```bash uv --directory "/path/to/your/project/host/src" run mcp_host_tutorial_openai.py ``` **Important**: Before running the above command, make sure to modify the MCP server configuration path in the agent file (the `--directory` argument in RAW_CONFIG) to match your environment. ## Common Issues and Solutions ### If the API Key is Not Recognized - Ensure that the `.env` file is correctly placed in the root directory of the project. - Check that there are no extra spaces or quotes in the API key within the file. ### If You Reach Google API Quota Limits - Charges will apply if you exceed the daily free quota (100 requests). - Check the quota and usage in the Google Cloud Console.