readme

Author: JakubS-Kentico

README.md

@@ -1 +1,78 @@
-# McpTesting
+# MCPBro - Gemini Chatbot with MCP Tools
+
+This project is a web-based chatbot powered by Google Gemini, utilizing the Model Context Protocol (MCP) to interact with external tools like filesystem access and memory.
+
+## Features
+
+*   **Web Interface:** Simple chat interface built with Flask and SocketIO.
+*   **Gemini Integration:** Uses the Google Generative AI SDK for chat responses.
+*   **MCP Tool Usage:** Connects to MCP servers (filesystem, memory) to provide tools to the Gemini model.
+*   **Real-time Communication:** Uses Flask-SocketIO for instant message updates.
+*   **Configuration:** Flexible configuration via `config.json` and environment variables.
+*   **Dependency Management:** Uses Poetry for managing Python dependencies.
+
+## Prerequisites
+
+*   Python >= 3.13
+*   [Poetry](https://python-poetry.org/docs/#installation)
+*   Node.js and npx (for running MCP servers like `@modelcontextprotocol/server-filesystem` and `@modelcontextprotocol/server-memory`)
+*   A Google Gemini API Key
+
+## Setup
+
+1.  **Clone the repository:**
+    ```bash
+    git clone <your-repository-url>
+    cd mcpbro
+    ```
+
+2.  **Install dependencies:**
+    ```bash
+    poetry install
+    ```
+    This will install all necessary Python packages based on the `poetry.lock` file.
+
+3.  **Configure the application:**
+    *   Copy the sample configuration file:
+        ```bash
+        cp bot_config/config.json.sample bot_config/config.json
+        ```
+    *   Edit `bot_config/config.json`:
+        *   Set `gemini_api_key` to your Google Gemini API key (alternatively, set the `GEMINI_API_KEY` environment variable).
+        *   Add the absolute paths to the directories you want the filesystem tool to access under `filesystem_target_directories`.
+        *   Set `enable_memory_server` to `true` if you want to use the memory tool.
+    *   (Optional) Edit `bot_config/system_instruction.md` to customize the chatbot's behavior and personality.
+
+## Running the Application
+
+1.  **Start the server:**
+    You can use the Poe the Poet task defined in `pyproject.toml`:
+    ```bash
+    poetry run poe start
+    ```
+    Alternatively, run directly:
+    ```bash
+    poetry run python run.py
+    ```
+
+2.  **Access the application:**
+    Open your web browser and navigate to `http://127.0.0.1:5000` (or the host/port specified in the console output).
+
+## Usage
+
+*   Type your messages into the input box and press Enter or click Send.
+*   The chatbot will respond, potentially using the configured MCP tools.
+*   Internal steps (like tool usage) will be displayed in the chat interface.
+*   Use the `/reset` endpoint (e.g., `http://127.0.0.1:5000/reset`) to clear the chat history.
+*   Use the `/debug` endpoint to view server-side debug logs.
+
+## Key Dependencies
+
+*   Flask
+*   Flask-SocketIO
+*   Flask-Session
+*   gevent
+*   google-generativeai
+*   mcp
+
+See `pyproject.toml` for the full list of dependencies.