ByteRover Extension

The ByteRover Extension enables persistent memory features for your AI coding assistants, including Windsurf, Cursor, and Cline. This extension simplifies the process of setting up the ByteRover Memory MCP (Memory Control Protocol) server and configuring it across your development environments.

Features

• Persistent Memory: Keep important context across coding sessions
• Automatic Memory Management: ByteRover automatically saves and retrieves relevant code context
• Multi-IDE Support: Configure once, use across Windsurf, Cursor, and Cline
• Custom AI Instructions: Create and deploy your own AI behavior instructions
• Simple Server Management: Start, monitor, and restart the MCP server directly from the VS Code sidebar

Requirements

• ByteRover account with a valid public API key

User Guide

Installation

1. Install the ByteRover Extension from the VS Code Marketplace or Open VSX Registry
2. Once installed, the extension will automatically add a ByteRover icon to your activity bar
3. Click on the ByteRover icon to view the server status and management options

Setup and Configuration

1. The extension will automatically install the ByteRover MCP server on first activation
2. download configuration file (byterover.conf) from ByteRover to your workspace
3. The extension will automatically deploy the configuration to supported IDEs

Using the Extension

Server Management

The ByteRover sidebar provides real-time status of your MCP server:

• View server status (running, starting, idle, or error)
• Restart the server when needed by clicking the "Restart MCP Server" option

Custom AI Instructions

ByteRover lets you define custom AI behavior instructions that apply across all supported IDEs:

1. Add your custom instructions to the configuration file
2. The extension automatically deploys these instructions to all detected IDEs
3. Default instructions encourage the AI to use memory tools for better context awareness:

# important 
always use search-memories tool to get the related context before any tasks 
always use create-memories to store all the critical informations after sucessfull tasks

IDE-Specific Configuration

ByteRover deploys configurations to the appropriate location for each IDE:

• Cursor: .cursor/mcp.json (project-level only)
• Windsurf: ~/.codeium/windsurf/mcp_config.json
• Cline: .cline/mcp.json

Note for Cursor users: Cursor only supports project-level configuration, not global configuration.

Verification

To verify ByteRover is working correctly:

1. Check the ByteRover sidebar to confirm the server is running
2. Open your AI coding assistant and confirm it can access memory features
3. Try creating a memory by asking your AI assistant to remember something important
4. In a later session, verify the AI can recall this information

Benefits of ByteRover Memory

Persistent Context Across Sessions

With ByteRover Memory enabled, your AI coding assistant can retain important context about your project even after you close your IDE or restart your computer. This means:

• No need to repeatedly explain your project structure
• AI remembers your coding preferences and project-specific details
• Faster, more relevant responses based on historical context

Team Collaboration

When working in a team environment, ByteRover helps maintain collective knowledge:

• Share project-specific AI instructions across your team
• Build a shared memory of important project details
• Ensure consistent AI behavior across different team members

Troubleshooting

Server Issues

If you encounter issues with the MCP server:

1. Check the server status in the ByteRover sidebar
2. Try restarting the server using the provided button
3. Ensure your system has the necessary permissions for the configuration files

Configuration Problems

If your AI assistant doesn't seem to use the memory features:

1. Verify your ByteRover public API key and user ID are correct
2. Ensure the configuration has been deployed to your IDE's configuration location
3. Check that the MCP server is running
4. Restart your AI coding assistant