Kilocode Configuration

Kilocode is a powerful VSCode extension that provides AI-powered code completion, inline chat, and code actions. Configure it to use MegaLLM for access to multiple AI models.

Quick Configuration

Via Settings UI
Via settings.json
Using Environment Variables

Step-by-Step Setup

Open VSCode Settings

Press Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (Mac)
Type: Preferences: Open Settings (UI)
Search for: Kilocode

Configure API Provider

API Provider: Select Custom
Provider Name: MegaLLM
Base URL: https://ai.megallm.io/v1
API Key: sk-mega-your-api-key-here

Select Default Model

Default Model: gpt-5 (or any supported model)
Temperature: 0.3 (lower = more deterministic)
Max Tokens: 500 (for completions)

Enable Features

Enable AutoComplete
Enable Inline Chat
Enable Code Actions
Enable Suggestions

Configuration File

Location: .vscode/settings.json or ~/.config/Code/User/settings.json

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "sk-mega-your-api-key-here"
  },
  "kilocode.defaultModel": "gpt-5",
  "kilocode.enableAutoComplete": true,
  "kilocode.enableInlineChat": true,
  "kilocode.enableCodeActions": true,
  "kilocode.modelSettings": {
    "temperature": 0.3,
    "maxTokens": 500,
    "topP": 0.9
  }
}

You can use environment variables: "apiKey": "${env:MEGALLM_API_KEY}"

Secure Configuration

Step 1: Set environment variables

# Add to ~/.bashrc, ~/.zshrc, or ~/.config/fish/config.fish
export MEGALLM_API_KEY="sk-mega-your-api-key-here"
export MEGALLM_BASE_URL="https://ai.megallm.io/v1"

Step 2: Reload shell

source ~/.bashrc  # or ~/.zshrc

Step 3: Reference in settings.json

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "${env:MEGALLM_BASE_URL}",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },
  "kilocode.defaultModel": "gpt-5"
}

API key stays secure and out of version control

Scenario Examples

Scenario 1: First-Time Installation

Complete setup from scratch:

Install Kilocode Extension

Open VSCode
Go to Extensions: Ctrl+Shift+X / Cmd+Shift+X
Search: Kilocode
Click Install
Reload VSCode window

Get MegaLLM API Key

Visit MegaLLM Dashboard
Navigate to API Keys section
Click Create New Key
Copy the key (starts with sk-mega-)
Store it securely

Configure Extension

Open settings (Ctrl+, / Cmd+,) and add:

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "sk-mega-your-actual-key-here"
  },
  "kilocode.defaultModel": "gpt-5"
}

Test Configuration

Create a new file: test.js
Type a comment: // function to calculate fibonacci
Press Tab to trigger completion
Should see AI-generated code

Expected Result:

// function to calculate fibonacci
function fibonacci(n) {
  if (n <= 1) return n;
  return fibonacci(n - 1) + fibonacci(n - 2);
}

Scenario 2: Team Project Configuration

Set up Kilocode for entire development team: Project Structure:

my-project/
├── .vscode/
│   ├── settings.json          # Shared config (committed)
│   └── settings.local.json    # Personal keys (gitignored)
├── .gitignore
└── README.md

Step 1: Create Shared Config .vscode/settings.json (committed to git):

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },
  "kilocode.defaultModel": "gpt-5",
  "kilocode.modelSettings": {
    "temperature": 0.7,
    "maxTokens": 1000
  },
  "kilocode.enableAutoComplete": true,
  "kilocode.enableInlineChat": true
}

Step 2: Setup .gitignore

# .gitignore
.vscode/settings.local.json
.env

Step 3: Create Setup Instructions README.md:

## Kilocode Setup

### Prerequisites
1. VSCode with Kilocode extension installed
2. MegaLLM API key ([Get one](https://megallm.io/dashboard))

### Configuration

1. **Set environment variable:**
   ```bash
   export MEGALLM_API_KEY="your-key-here"

Add to your shell config (~/.bashrc or ~/.zshrc) to persist.

Or create local settings (not committed): .vscode/settings.local.json:

{
  "kilocode.customProvider": {
    "apiKey": "your-key-here"
  }
}

Reload VSCode and start coding!

Verification

Type // hello world function and press Tab. Should see AI-generated code.

**Step 4: Team Members Clone and Setup**

```bash
# Team member workflow
git clone https://github.com/company/my-project.git
cd my-project

# Add personal API key
export MEGALLM_API_KEY="their-personal-key"

# Or create .vscode/settings.local.json with their key

# Open in VSCode
code .

Scenario 3: Project-Specific Model Selection

Use different models for different projects: Python Data Science Project: .vscode/settings.json:

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },
  "kilocode.defaultModel": "claude-opus-4-1-20250805",
  "kilocode.customPrompts": {
    "analyze": "Analyze this data processing code for efficiency",
    "doc": "Generate numpy-style docstring for this function"
  }
}

JavaScript/React Project: .vscode/settings.json:

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },
  "kilocode.defaultModel": "gpt-5",
  "kilocode.customPrompts": {
    "component": "Generate a React functional component",
    "test": "Write Jest tests for this component"
  }
}

Systems Programming (Rust/Go): .vscode/settings.json:

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },
  "kilocode.defaultModel": "gemini-2.5-pro",
  "kilocode.modelSettings": {
    "temperature": 0.2  // Lower for more precise systems code
  }
}

Scenario 4: Multi-Model Workflow

Switch models dynamically based on task: Configuration:

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },
  "kilocode.defaultModel": "gpt-5",
  "kilocode.modelProfiles": {
    "fast": {
      "model": "gpt-4o-mini",
      "temperature": 0.5,
      "maxTokens": 300
    },
    "quality": {
      "model": "claude-opus-4-1-20250805",
      "temperature": 0.3,
      "maxTokens": 1000
    },
    "creative": {
      "model": "gpt-5",
      "temperature": 0.9,
      "maxTokens": 800
    }
  }
}

Usage:

Morning (rapid prototyping): Use fast profile with GPT-4o-mini
Afternoon (quality code): Use quality profile with Claude Opus
Documentation: Use creative profile with higher temperature

Switch profiles via Command Palette:

Ctrl+Shift+P / Cmd+Shift+P
Kilocode: Switch Model Profile
Select: fast, quality, or creative

Scenario 5: Migration from GitHub Copilot

Switching from Copilot to Kilocode with MegaLLM: Current Setup (Copilot):

{
  "github.copilot.enable": true
}

New Setup (Kilocode + MegaLLM):

Disable Copilot

{
  "github.copilot.enable": false
}

Install Kilocode

VSCode Extensions → Search “Kilocode” → Install

Configure MegaLLM

{
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },
  "kilocode.defaultModel": "gpt-5"
}

Compare Experience

Benefits over Copilot:

Access to multiple models (GPT, Claude, Gemini)
Better pricing and no seat limits
Inline chat for explanations
Custom model selection per project
Code actions beyond completion

Configuration Options

Complete Reference

{
  // Provider Configuration
  "kilocode.apiProvider": "custom",
  "kilocode.customProvider": {
    "name": "MegaLLM",
    "baseURL": "https://ai.megallm.io/v1",
    "apiKey": "${env:MEGALLM_API_KEY}"
  },

  // Model Settings
  "kilocode.defaultModel": "gpt-5",
  "kilocode.modelSettings": {
    "temperature": 0.3,        // 0.0-2.0: Lower = more focused
    "maxTokens": 500,          // Max completion length
    "topP": 0.9,               // Nucleus sampling
    "frequencyPenalty": 0.0,   // Reduce repetition
    "presencePenalty": 0.0     // Encourage diversity
  },

  // Feature Toggles
  "kilocode.enableAutoComplete": true,
  "kilocode.enableInlineChat": true,
  "kilocode.enableCodeActions": true,
  "kilocode.enableSuggestions": true,
  "kilocode.enableHover": true,

  // Behavior
  "kilocode.autoTrigger": true,
  "kilocode.debounceDelay": 300,     // ms before triggering
  "kilocode.maxSuggestions": 3,
  "kilocode.showInlineHints": true,

  // Custom Prompts
  "kilocode.customPrompts": {
    "doc": "Generate comprehensive documentation",
    "test": "Write unit tests with high coverage",
    "refactor": "Refactor for better maintainability",
    "optimize": "Optimize for performance",
    "secure": "Review for security vulnerabilities"
  },

  // Keyboard Shortcuts
  "kilocode.shortcuts": {
    "acceptSuggestion": "Tab",
    "nextSuggestion": "Alt+]",
    "prevSuggestion": "Alt+[",
    "dismissSuggestion": "Esc",
    "triggerInlineChat": "Ctrl+K"
  },

  // UI Preferences
  "kilocode.ui": {
    "showStatusBar": true,
    "showInlineButtons": true,
    "theme": "auto",               // auto, light, dark
    "position": "right"            // left, right
  }
}

Model Selection Guide

Task	Recommended Model	Reason
Code Completion	`gpt-4o-mini`	Fast, cost-effective
Complex Logic	`claude-opus-4-1-20250805`	Superior reasoning
Web Development	`gpt-5`	Excellent JS/TS/React
Data Science	`claude-sonnet-4`	Strong analysis
Documentation	`gpt-5`	Clear explanations
Algorithms	`gemini-2.5-pro`	Mathematical precision

See full model catalog for all options.

Verification

Test 1: Basic Completion

// Type this comment and press Tab:
// function to check if string is palindrome

// Expected output:
function isPalindrome(str) {
  const cleaned = str.toLowerCase().replace(/[^a-z0-9]/g, '');
  return cleaned === cleaned.split('').reverse().join('');
}

Test 2: Inline Chat

Select a function
Press Ctrl+K (or Cmd+K on Mac)
Type: Explain this function
Should see explanation in chat panel

Test 3: Code Actions

Right-click on code
Should see “Kilocode Actions” in context menu
Options: Explain, Improve, Generate Tests, etc.

Test 4: Status Bar

Check bottom-right of VSCode:

Should show: Kilocode: Connected
Model name: gpt-5 (or your selected model)
Click to see connection details

Troubleshooting

Completions not appearing

Symptoms:

No suggestions when typing
Status bar shows “Disconnected”

Solutions:

Check API key:

echo $MEGALLM_API_KEY
# Should output: sk-mega-...

Verify configuration:

{
  "kilocode.apiProvider": "custom",  // Must be "custom"
  "kilocode.customProvider": {
    "baseURL": "https://ai.megallm.io/v1"  // No trailing slash
  }
}

Reload VSCode:
- Ctrl+Shift+P / Cmd+Shift+P
- Run: Developer: Reload Window
Check extension is enabled:
- Extensions panel
- Search: Kilocode
- Should show “Enabled”

Test API manually:

curl -X POST https://ai.megallm.io/v1/chat/completions \
  -H "Authorization: Bearer $MEGALLM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"test"}]}'

Wrong or poor quality completions

Symptoms:

Completions are incorrect
Suggestions don’t match code style
Irrelevant responses

Solutions:

Adjust temperature:

{
  "kilocode.modelSettings": {
    "temperature": 0.2  // Lower = more focused (0.0-1.0)
  }
}

Try different model:

{
  "kilocode.defaultModel": "claude-sonnet-4"  // Better for analysis
}

Increase context:

{
  "kilocode.modelSettings": {
    "maxTokens": 1000  // More detailed completions
  }
}

Add project-specific prompt:

{
  "kilocode.systemPrompt": "You are an expert in TypeScript and React. Follow Airbnb style guide."
}

High latency / slow completions

Symptoms:

Long wait for suggestions
Timeout errors

Solutions:

Use faster model:

{
  "kilocode.defaultModel": "gpt-4o-mini"  // Fastest
}

Reduce max tokens:

{
  "kilocode.modelSettings": {
    "maxTokens": 300  // Faster completions
  }
}

Increase debounce delay:

{
  "kilocode.debounceDelay": 500  // Wait longer before triggering
}

Check network:

ping ai.megallm.io
curl -w "@-" -o /dev/null -s https://ai.megallm.io/v1/models <<'EOF'
time_total: %{time_total}s
EOF

API key not recognized

Symptoms:

“Invalid API key” error
401 Unauthorized

Solutions:

Verify key format:
- Must start with sk-mega-
- At least 60 characters
- No spaces or quotes
Check key is active:
- Login to Dashboard
- Go to API Keys
- Verify key is not revoked/expired

Test key directly:

curl -H "Authorization: Bearer sk-mega-your-key" \
     https://ai.megallm.io/v1/models

Regenerate if needed:
- Dashboard → API Keys → Create New
- Update configuration with new key

Conflicts with other extensions

Symptoms:

Kilocode completions conflict with other AI tools
Multiple suggestions appearing

Solutions:

Disable conflicting extensions:

{
  "github.copilot.enable": false,
  "tabnine.disable": true,
  "aws.codeWhisperer.enabled": false
}

Adjust trigger keys:

{
  "kilocode.shortcuts": {
    "acceptSuggestion": "Alt+Enter"  // Different from Tab
  }
}

Set priority:

{
  "editor.suggest.snippetsPreventQuickSuggestions": false,
  "editor.quickSuggestions": {
    "other": "on",
    "comments": "off",
    "strings": "on"
  }
}

Best Practices

Use Environment Variables

Keep API keys in env vars, reference with ${env:MEGALLM_API_KEY}

Project-Specific Models

Configure different models in .vscode/settings.json per project

Lower Temperature for Code

Use 0.2-0.4 for code generation, 0.7-0.9 for documentation

Monitor Token Usage

Check Dashboard to optimize costs

Advanced Tips

Custom Keyboard Shortcuts

Add to keybindings.json:

[
  {
    "key": "ctrl+alt+k",
    "command": "kilocode.triggerSuggestion",
    "when": "editorTextFocus"
  },
  {
    "key": "ctrl+alt+e",
    "command": "kilocode.explainCode",
    "when": "editorHasSelection"
  },
  {
    "key": "ctrl+alt+t",
    "command": "kilocode.generateTests",
    "when": "editorTextFocus"
  }
]

Workspace-Specific Prompts

.vscode/settings.json:

{
  "kilocode.customPrompts": {
    "api": "Generate RESTful API endpoint following our conventions",
    "schema": "Create Prisma schema model",
    "hook": "Generate React custom hook",
    "component": "Create styled-component with TypeScript"
  }
}

Context-Aware Completions

{
  "kilocode.contextSettings": {
    "includeOpenFiles": true,
    "includeImports": true,
    "includeTypes": true,
    "maxContextFiles": 5
  }
}

Next Steps

RooCode Setup

Configure RooCode application

Cline Setup

Configure Cline VSCode extension

Models Catalog

Browse all available models

Other Agents

View all CLI and GUI agents

API Reference

Explore the API docs

Discord Community

Join for support

Overview

Configuration

Documentation Index

​Quick Configuration

​Step-by-Step Setup

​Configuration File

​Secure Configuration

​Scenario Examples

​Scenario 1: First-Time Installation

​Scenario 2: Team Project Configuration

​Verification

​Scenario 3: Project-Specific Model Selection

​Scenario 4: Multi-Model Workflow

​Scenario 5: Migration from GitHub Copilot

​Configuration Options

​Complete Reference

​Model Selection Guide

​Verification

​Test 1: Basic Completion

​Test 2: Inline Chat

​Test 3: Code Actions

​Test 4: Status Bar

​Troubleshooting

​Best Practices

Use Environment Variables

Project-Specific Models

Lower Temperature for Code

Monitor Token Usage

​Advanced Tips

​Custom Keyboard Shortcuts

​Workspace-Specific Prompts

​Context-Aware Completions

​Next Steps

RooCode Setup

Cline Setup

Models Catalog

Other Agents

API Reference

Discord Community

Quick Configuration

Step-by-Step Setup

Configuration File

Secure Configuration

Scenario Examples

Scenario 1: First-Time Installation

Scenario 2: Team Project Configuration

Verification

Scenario 3: Project-Specific Model Selection

Scenario 4: Multi-Model Workflow

Scenario 5: Migration from GitHub Copilot

Configuration Options

Complete Reference

Model Selection Guide

Verification

Test 1: Basic Completion

Test 2: Inline Chat

Test 3: Code Actions

Test 4: Status Bar

Troubleshooting

Best Practices

Advanced Tips

Custom Keyboard Shortcuts

Workspace-Specific Prompts

Context-Aware Completions

Next Steps