Command Palette

Search for a command to run...

GitHubGitHub

VS Code AI Extensions Installation Guide

Add AI to Visual Studio Code with GitHub Copilot, Cline, Roo Code and Continue, install them from the command line and configure them for a team.

You do not have to switch editors to get an AI assistant. The VS Code extension marketplace has everything from inline completion to fully autonomous agents.

Installing the Extensions

From the Command Line

# GitHub Copilot — inline completion and chat
code --install-extension GitHub.copilot
code --install-extension GitHub.copilot-chat
 
# Cline — an autonomous agent that edits files and runs commands
code --install-extension saoudrizwan.claude-dev
 
# Roo Code — the community fork of Cline with extra modes
code --install-extension RooVeterinaryInc.roo-cline
 
# Continue — an open-source assistant and agent
code --install-extension Continue.continue
 
# List what is installed
code --list-extensions

Choosing an Extension

ExtensionModelBest for
GitHub CopilotGitHub's ownInline completion, chat, and an agent mode inside VS Code
ClineBring your own keyA fully autonomous agent that plans, edits and runs commands
Roo CodeBring your own keyThe same, with extra modes and a smaller footprint
ContinueBring your own keyOpen-source, configurable, works with local models

GitHub Copilot

# Sign in
code --install-extension GitHub.copilot

Sign in from the Accounts icon in the activity bar, or with the GitHub CLI:

gh auth login
// .vscode/settings.json — repository level defaults
{
  "github.copilot.chat.codeGeneration.useInstructionFiles": true
}

Put your repository instructions in .github/copilot-instructions.md and commit them.

Cline and Roo Code

Both agents work the same way: you pick a model provider, paste an API key, and describe the task. They read the project, propose edits and run commands in a terminal.

// .vscode/settings.json — keep the agent out of sensitive paths
{
  "cline.deniedCommands": ["rm -rf", "git push --force"],
  "cline.customInstructions": "Run `npm test` before you say a change is done."
}

Continue

Continue is fully open source and works with any provider, including local models:

# ~/.continue/config.yaml
name: My Assistant
version: 1.0.0
schema: v1
models:
  - name: Claude Sonnet
    provider: anthropic
    model: claude-sonnet-4-5
    apiKey: $ANTHROPIC_API_KEY
  - name: Llama (local)
    provider: ollama
    model: llama3.1
    apiBase: http://localhost:11434/v1
# The models tab in the sidebar
code --install-extension Continue.continue

Troubleshooting

Best Practices

  1. Pick one agent and keep the rest for completion only.
  2. Commit the instructions file — .github/copilot-instructions.md or .continue/rules — so the team gets the same behaviour.
  3. Run agents in a dev container so they cannot touch your machine.
  4. Review every diff and read the terminal output the agent produced.
  5. Keep API keys in the OS keychain or a secret store, never in settings.json that you commit.
  6. Pin the extension versions in the dev container image for reproducible setups.
  7. Add the generated files to .gitignore — agents create scratch files.

Next Steps

OpenCode

Install the open-source OpenCode terminal agent with the install script, npm or Homebrew, connect a model provider and bootstrap the project with an AGENTS.md file.

Overview

Install and configure Docker, Kubernetes, Helm, Istio, Minikube and OpenShift with step-by-step guides, production-ready configuration examples and troubleshooting tips.