VS Code Coding Assistants

Overview

Connect the gateway to popular VS Code AI coding assistants for autonomous coding with TEE privacy protection. All three assistants support APIs.

Kilo Code

Fork of Roo/Cline with 400+ models

Continue.dev

Ship faster with continuous AI

Cline

Autonomous coding agent

Kilo Code

Kilo Code is an open source AI coding assistant for planning, building, and fixing code.

Installation

Open VS Code
Go to Extensions (Cmd+Shift+X or Ctrl+Shift+X)
Search for “Kilo Code”
Click Install

Configuration

Open Kilo Code Settings

Click the Kilo Code icon in the sidebar → Click ⚙️ Settings

Select API Provider

Choose: OpenAI Compatible

Configure API

Base URL: https://api.redpill.ai/v1
API Key: YOUR_API_KEY

Select Model

Choose from the available models:

z-ai/glm-5.1 - Best confidential reasoning
z-ai/glm-5 - Best confidential systems engineering
qwen/qwen3.5-27b - Runs on a verified provider
openai/gpt-5 - Best general purpose
anthropic/claude-sonnet-4.5 - Best reasoning
deepseek/deepseek-chat-v3.1 - Best for coding

Configuration JSON

{
  "kilocode.apiProvider": "openai-compatible",
  "kilocode.openaiCompatible.baseUrl": "https://api.redpill.ai/v1",
  "kilocode.openaiCompatible.apiKey": "YOUR_API_KEY",
  "kilocode.openaiCompatible.model": "deepseek/deepseek-chat-v3.1"
}

Using Kilo Code

@kilo create a REST API endpoint for user authentication

Kilo Code Features

✅ Autonomous Planning - AI plans multi-step tasks
✅ File Operations - Create, edit, delete files
✅ Terminal Commands - Execute commands with permission
✅ Code Review - Analyze and improve code
✅ Multi-File Editing - Work across entire codebase
✅ 400+ Models - Switch models anytime

Recommended Models for Kilo Code

Task	Model	Why
Sensitive Code	`z-ai/glm-5.1`	GPU TEE protection
General Coding	`deepseek/deepseek-chat-v3.1`	Optimized for code
Complex Refactoring	`anthropic/claude-sonnet-4.5`	Best reasoning
Quick Fixes	`google/gemini-2.5-flash`	Fast responses

Continue.dev

Continue.dev is an open-source AI code assistant for VS Code and JetBrains.

Installation

Open VS Code
Go to Extensions
Search for “Continue”
Click Install

Configuration

Open Config File

Press Cmd+Shift+P (Mac) or Ctrl+Shift+P (Windows/Linux)Type: Continue: Open config.json

Add the Configuration

Add to the models array:

{
  "models": [
    {
      "title": "GLM 5.1 TEE",
      "provider": "openai",
      "model": "z-ai/glm-5.1",
      "apiKey": "YOUR_API_KEY",
      "apiBase": "https://api.redpill.ai/v1"
    },
    {
      "title": "GLM-5 TEE",
      "provider": "openai",
      "model": "z-ai/glm-5",
      "apiKey": "YOUR_API_KEY",
      "apiBase": "https://api.redpill.ai/v1"
    },
    {
      "title": "GPT-5",
      "provider": "openai",
      "model": "openai/gpt-5",
      "apiKey": "YOUR_API_KEY",
      "apiBase": "https://api.redpill.ai/v1"
    },
    {
      "title": "Claude 4.5",
      "provider": "openai",
      "model": "anthropic/claude-sonnet-4.5",
      "apiKey": "YOUR_API_KEY",
      "apiBase": "https://api.redpill.ai/v1"
    },
    {
      "title": "DeepSeek",
      "provider": "openai",
      "model": "deepseek/deepseek-chat-v3.1",
      "apiKey": "YOUR_API_KEY",
      "apiBase": "https://api.redpill.ai/v1"
    }
  ]
}

Save and Reload

Save the config file. Continue will auto-reload.

Complete config.json Example

{
  "models": [
    {
      "title": "GLM 5.1 TEE",
      "provider": "openai",
      "model": "z-ai/glm-5.1",
      "apiKey": "YOUR_API_KEY",
      "apiBase": "https://api.redpill.ai/v1",
      "systemMessage": "You are an expert software developer."
    },
    {
      "title": "GPT-5",
      "provider": "openai",
      "model": "openai/gpt-5",
      "apiKey": "YOUR_API_KEY",
      "apiBase": "https://api.redpill.ai/v1",
      "systemMessage": "You are an expert software developer."
    }
  ],
  "embeddingsProvider": {
    "provider": "openai",
    "model": "openai/text-embedding-3-small",
    "apiKey": "YOUR_API_KEY",
    "apiBase": "https://api.redpill.ai/v1"
  },
  "tabAutocompleteModel": {
    "title": "DeepSeek",
    "provider": "openai",
    "model": "deepseek/deepseek-chat-v3.1",
    "apiKey": "YOUR_API_KEY",
    "apiBase": "https://api.redpill.ai/v1"
  }
}

Using Continue.dev

Cmd+L (Mac) or Ctrl+L (Windows/Linux)
Type your question or request

Continue.dev Features

✅ Chat - Ask questions about your code
✅ Edit - Modify code with AI assistance
✅ Autocomplete - Tab completion powered by AI
✅ Context - Understands your entire codebase
✅ Multi-Model - Switch models per task
✅ Custom Commands - Create slash commands

Cline

Cline is an autonomous coding agent that can create/edit files, execute commands, and use the browser.

Installation

Open VS Code
Go to Extensions
Search for “Cline”
Click Install

Configuration

Open Cline

Click the Cline robot icon in the sidebar

Click Settings

Click ⚙️ in the top right corner

Select API Provider

Choose: OpenAI Compatible

Enter Details

Base URL: https://api.redpill.ai/v1
API Key: YOUR_API_KEY
Model ID: Enter model name (e.g., openai/gpt-5)

Configuration via Settings UI

API Provider: OpenAI Compatible Base URL:

https://api.redpill.ai/v1

API Key:

YOUR_API_KEY

Model ID (examples):

z-ai/glm-5.1
z-ai/glm-5
qwen/qwen3.5-27b
openai/gpt-5
anthropic/claude-sonnet-4.5
deepseek/deepseek-chat-v3.1

Using Cline

Cline is task-oriented and autonomous:

"Create a user authentication system with JWT tokens, including login, logout, and middleware"

Cline Features

✅ Autonomous - Plans and executes multi-step tasks
✅ File Operations - Create, edit, delete files
✅ Terminal - Run commands with approval
✅ Browser - Use browser for research
✅ Approval Flow - Review before executing
✅ Token Tracking - Monitor API usage

Approval System

Cline asks permission before:

Creating/editing files
Running terminal commands
Making API calls
Using browser

Model Recommendations by Task

Task	Best Model	Alternative
Sensitive Code	`z-ai/glm-5.1`	`qwen/qwen3.5-27b`
New Features	`anthropic/claude-sonnet-4.5`	`openai/gpt-5`
Bug Fixes	`deepseek/deepseek-chat-v3.1`	`openai/gpt-5`
Refactoring	`anthropic/claude-sonnet-4.5`	`deepseek/deepseek-chat-v3.1`
Documentation	`openai/gpt-5`	`google/gemini-2.5-flash`
Code Review	`anthropic/claude-sonnet-4.5`	`openai/gpt-5`
Quick Edits	`google/gemini-2.5-flash`	`deepseek/deepseek-chat-v3.1`

Comparison

Feature	Kilo Code	Continue.dev	Cline
Autonomous Planning	✅ Yes	⚠️ Limited	✅ Yes
Chat Interface	✅ Yes	✅ Yes	✅ Yes
Tab Autocomplete	❌ No	✅ Yes	❌ No
File Operations	✅ Yes	✅ Yes	✅ Yes
Terminal Access	✅ Yes	✅ Yes	✅ Yes
Browser Use	❌ No	❌ No	✅ Yes
Approval Flow	✅ Yes	⚠️ Partial	✅ Yes
Multi-Model	✅ 400+	✅ Custom	✅ Custom
Cost Tracking	✅ Yes	❌ No	✅ Yes

Best Practices

Choose the Right Model

GPU TEE models: Best for proprietary code
GLM 5.1: Best for confidential reasoning
GLM-5: Best for confidential systems engineering
DeepSeek: Best for pure coding tasks
Claude 4.5: Best for complex reasoning and refactoring
GPT-5: Best for general purpose
Gemini Flash: Best for speed

Provide Clear Context

Describe the full task upfront
Include relevant file paths
Mention tech stack and dependencies
Specify coding style preferences

Review AI Changes

Always review code before committing
Test generated code thoroughly
Check for security issues
Verify error handling

Manage Costs

Use cheaper models for simple tasks
Set token limits in settings
Monitor usage in the dashboard
Use Continue autocomplete sparingly

Privacy for Proprietary Code

Requests are served through the attested TEE gateway, which does not retain request bodies
Use confidential (is_tee) models for verified-upstream execution
Enable local caching in Continue config

Environment Variables

Store API keys securely:

.env

API_KEY=YOUR_API_KEY

Reference in configs:

{
  "apiKey": "${API_KEY}"
}

Troubleshooting

Connection Errors

Solution:

Verify base URL: https://api.redpill.ai/v1
Check API key is valid
Ensure model name includes provider prefix
Reload VS Code window

Model Not Found

Solution:

Use full model name: openai/gpt-5 not gpt-5
Check model exists in supported models
Try a different model

Slow Responses

Solution:

Switch to faster model (Gemini Flash, DeepSeek)
Reduce context window
Check internet connection
Clear VS Code cache

High API Costs

Solution:

Use cheaper models for simple tasks
Disable autocomplete if using Continue
Set max_tokens limits
Monitor usage in the dashboard

Example Workflows

Workflow 1: Feature Development

Kilo Code: “Plan a user authentication system”
Review the plan
Kilo Code: “Implement the plan”
Continue: Ask questions about specific parts
Cline: “Add tests for the auth system”

Workflow 2: Bug Fixing

Continue Chat: “What’s causing this error?”
Review explanation
Cline: “Fix the bug and add error handling”
Continue Edit: Make final adjustments

Workflow 3: Code Review

Kilo Code: “Review this PR for issues”
Continue: “Explain this complex function”
Cline: “Refactor based on review feedback”

Resources

Next Steps

Supported Models

Browse all available models

Function Calling

Learn advanced AI features

Privacy Features

Understand TEE protection

Confidential Models

Use confidential AI models

​Overview

Kilo Code

Continue.dev

Cline

​Kilo Code

​Installation

​Configuration

​Configuration JSON

​Using Kilo Code

​Kilo Code Features

​Recommended Models for Kilo Code

​Continue.dev

​Installation

​Configuration

​Complete config.json Example

​Using Continue.dev

​Continue.dev Features

​Cline

​Installation

​Configuration

​Configuration via Settings UI

​Using Cline

​Cline Features

​Approval System

​Model Recommendations by Task

​Comparison

​Best Practices

​Environment Variables

​Troubleshooting

​Example Workflows

​Workflow 1: Feature Development

​Workflow 2: Bug Fixing

​Workflow 3: Code Review

​Resources

​Next Steps

Supported Models

Function Calling

Privacy Features

Confidential Models

Overview

Kilo Code

Installation

Configuration

Configuration JSON

Using Kilo Code

Kilo Code Features

Recommended Models for Kilo Code

Continue.dev

Installation

Configuration

Complete config.json Example

Using Continue.dev

Continue.dev Features

Cline

Installation

Configuration

Configuration via Settings UI

Using Cline

Cline Features

Approval System

Model Recommendations by Task

Comparison

Best Practices

Environment Variables

Troubleshooting

Example Workflows

Workflow 1: Feature Development

Workflow 2: Bug Fixing

Workflow 3: Code Review

Resources

Next Steps