ZeroCode Agent 🧠

ZeroCode is your AI-first coding assistant right inside VS Code. It brings an autonomous agent directly into your editor, letting you write, edit, and test code using natural language.

"Your codebase as a conversation."

What Can ZeroCode Do?

The ZeroCode Agent is an autonomous system that can navigate your workspace, read files, and write code on your behalf.

• Project Context: The agent automatically reads the files you are currently looking at. If you ask a question about your code, it already knows what you're working on!
• Cmd+K Edits: Highlight a block of code, press Cmd+K (or Ctrl+K), and type a prompt. The agent will rewrite the code in place!
• Multi-File Refactors: You can ask the agent in the sidebar to rewrite entire systems across multiple files, and it will execute the commands in the background.
• Customizable Skills: Drop a .md file into .zcode/skills and the agent will absorb it into its brain, perfectly adapting to your team's coding conventions.

Getting Started & Configuration

Configuring ZeroCode is easy. You can configure it to use your preferred AI model. Here is how:

1. Click the >_ icon in the left Activity Bar to open the Agent Chat, or press Ctrl/Cmd + Shift + P and type ZeroCode: Open Settings.
2. Navigate to your VS Code Settings (Ctrl/Cmd + ,).
3. Search for ZeroCode Agent to access the configuration panel.

4. Select your preferred Provider (e.g., gemini, openrouter, openai, anthropic, groq).
5. Enter the model name in the Model field (e.g., gemini-1.5-pro or meta-llama/llama-3-8b-instruct:free).
6. Paste your API Key in the corresponding field for your provider.

Note: The agent will automatically restart and apply your new settings immediately when you make changes!

How to get a Free API Key

You don't need to pay to use powerful AI models! Here are two great ways to get started for free:

Google AI Studio (Gemini)

Google provides generous free tiers for their state-of-the-art Gemini models.

1. Go to Google AI Studio.
2. Sign in with your Google account.
3. Click on Create API key and generate a new key.
4. In VS Code Settings, set the Provider to gemini, Model to gemini-1.5-pro, and paste your key into Gemini API Key.

OpenRouter (Llama 3 & Open Source Models)

OpenRouter provides access to over 100 free open-source models, including Meta's Llama 3.

1. Go to OpenRouter.
2. Sign up or log in.
3. Click on your profile in the top right, go to Keys, and click Create Key.
4. To find free models, go to the Chat page, click the model selector, and filter by "Free".
5. In VS Code Settings, set the Provider to openrouter, type the exact model ID (like meta-llama/llama-3-8b-instruct:free) into the Model field, and paste your key into OpenRouter API Key.

Supported Models

ZeroCode brings native support for every major LLM provider:

• OpenAI (gpt-4o, gpt-4-turbo)
• Anthropic (claude-3-5-sonnet-20240620)
• Google Gemini (gemini-1.5-pro)
• OpenRouter & Groq
• Local Models via Ollama/LMStudio

Just set your API key in the VS Code Settings and you are ready to go!

Model Context Protocol (MCP)

ZeroCode fully supports the Model Context Protocol (MCP) to connect your agent to external tools and databases (like PostgreSQL, GitHub, etc.).

Because MCP configurations can be complex, they are managed via a JSON config file.

1. Create a JSON config file anywhere on your computer (e.g. zcode_config.json) containing your mcp_servers block.
2. In VS Code, go to Settings and search for zerocode.agent.configPath.
3. Paste the absolute path to your JSON file. The agent will automatically load your MCP tools on startup!

Built for Vib Coding.