How to Use Gemini 2.5 (Free) to Code in Visual Studio Code скачать в хорошем качестве

How to Use Gemini 2.5 (Free) to Code in Visual Studio Code

Join my community at https://singulari.dev to continue learning about Artificial Intelligence. Discover how to set up your own programming assistant completely FREE using the power of Visual Studio Code, the Roo Code extension, and Google's powerful Gemini 2.5 model. In this step-by-step tutorial, we'll show you how to stop spending money on expensive tools and take advantage of a free yet incredibly capable setup. You'll learn how to: ✅ Install and configure Visual Studio Code on Windows. ✅ Install the Roo Code extension, a standalone AI agent for your editor. ✅ Connect Roo Code to large language models (LLMs) like Gemini 2.5 through OpenRouter or directly with Google AI Studio. ✅ Generate and use API keys securely to enable communication. ✅ Select free models like Gemini 2.5 Pro Experimental. ✅ Interact with the AI ​​agent in different modes (Architect, Code) to automatically plan and generate code. ✅ Create a functional web application (a beauty salon calculator) from scratch, simply describing the requirements in natural language. ✅ Understand key concepts such as tokens, context windows, and how to iterate with AI to refine the result. Chapters: 00:00 - Introduction: Mention of paid AI tools (Cursor, Lovable, Aider, GitHub Copilot) and their cost. 00:09 - Proposal: Save money with a free alternative. 00:46 - Free Solution: Visual Studio Code + Roo Code Extension + Free Gemini 2.5 Model. 01:09 - Community Mention: Invitation to singulari.dev. 01:16 - Starting Setup: Transition to Windows Desktop. 2:17 - Downloading Visual Studio Code: Search on Google and download from the official website. 2:30 - Installing Visual Studio Code: Running the installer and basic steps. 2:37 - VS Code Installation Options: Recommendation to enable contextual options and PATH. 2:50 - Optional Mention: VS Code Insiders. 3:16 - Launching VS Code: Initial launch of VS Code and welcome screen. 3:50 - Searching for the Roo Code Extension: Accessing the Extensions tab and searching. 4:00 - Installing Roo Code: Selecting and installing the extension (previously Roo Code). 4:13 - Roo Code Interface and Architecture: Brief overview of the interface and graphical explanation (VS Code - Roo Code - LLM Online). 5:04 - LLM Connection Options: Graphical explanation of OpenRouter and direct connection to Gemini (AI Studio). 5:37 - Connection via API Keys: Mention of how to connect. 6:12 - Connecting to OpenRouter (Part 1): Click the OpenRouter button in VS Code. 6:29 - OpenRouter Login: Enter the OpenRouter account in the browser. 7:00 - OpenRouter Authorization: Authorize the connection and return to VS Code. 7:19 - Model Configuration in Roo Code: Access settings and select the model (initially Claude 3.7 Sonnet). 7:46 - Selecting a Free Model (Gemini 2.5 via OpenRouter): Switch to the free model google-gemini-2.5-pro-exp-03-25:free and save settings. 8:30 AM - Roo Code Modes: Mention of the modes (Code, Architect, Ask, Debug). 8:55 AM - Example Prompt (Architect Mode): Pasting a detailed prompt for a beauty salon calculator. 10:55 AM - Model Response (Planning): The model responds with questions to clarify the plan. 11:12 AM - Clarifying Questions for the Model: Answering the model's questions to refine the plan. 1:04 PM - Generating the Detailed Plan: The model generates the plan and asks if it should be saved to a Markdown file. 4:30 PM - Saving the Plan: Confirmation to save the plan; Roo Code creates the .md file. 5:30 PM - Request to Switch to "Code" Mode: The model requests permission to switch to code mode and deploy. 6:21 PM - Starting the Deployment (Code Mode): Approval of the change; Roo Code begins generating the index.html, style.css, and script.js files. 8:00 PM - Error Correction: The model detects and corrects syntax errors in script.js. 8:10 PM - Run Application: The model suggests a command to open index.html. 8:14 PM - Application Demonstration: The generated web calculator is displayed in the browser and its functionality is tested. 9:26 PM - Feedback and Iterative Improvement: The model is informed that the app is incomplete and has errors in the console. 10:29 PM - Final Code Correction: The model corrects the script.js file again. 11:30 PM - Tokens and Context Window Explained: The use of tokens is demonstrated, and how the context window works is explained. 12:02 PM - Final Conclusion and Call to Action: A summary of the power of AI agents and a new invitation to the singulari.dev community.