Codex

Use XiaoYuan AI with Codex, the AI coding assistant.

Prerequisites

• Codex installed
• API key from https://ai.xiaoyuan.co.uk

Configuration

Codex supports OpenAI-compatible API endpoints, making it easy to use with XiaoYuan AI.

Step 1: Open Codex Settings

Access Codex settings through:

• Menu: Preferences → Settings
• Keyboard: Cmd/Ctrl + ,
• Command palette: Search for "Settings"

Step 2: Configure API Settings

Find the API configuration section and set:

Setting	Value
API Provider	OpenAI Compatible / Custom
Base URL	https://ai.xiaoyuan.co.uk/v1
API Key	Your XiaoYuan API key
Model	claude-opus-4-8

Configuration File

Alternatively, edit Codex's configuration file directly:

Linux/Mac: ~/.config/codex/config.json

{
  "api": {
    "provider": "openai-compatible",
    "baseUrl": "https://ai.xiaoyuan.co.uk/v1",
    "apiKey": "your-api-key-here",
    "model": "claude-opus-4-8"
  }
}

Windows: %APPDATA%/codex/config.json

{
  "api": {
    "provider": "openai-compatible",
    "baseUrl": "https://ai.xiaoyuan.co.uk/v1",
    "apiKey": "your-api-key-here",
    "model": "claude-opus-4-8"
  }
}

Using Environment Variables

For better security, use environment variables:

export CODEX_API_KEY="your-xiaoyuan-api-key"
export CODEX_BASE_URL="https://ai.xiaoyuan.co.uk/v1"
export CODEX_MODEL="claude-opus-4-8"

Then in config:

{
  "api": {
    "provider": "openai-compatible",
    "baseUrl": "${CODEX_BASE_URL}",
    "apiKey": "${CODEX_API_KEY}",
    "model": "${CODEX_MODEL}"
  }
}

Usage

Once configured, Codex will use Claude 4.8 for all AI interactions:

Chat Interface

Ask Claude anything about your code...

Code Generation

Use Codex commands to generate code:

• Cmd/Ctrl + K - Generate code
• Cmd/Ctrl + I - Inline suggestions
• Cmd/Ctrl + Shift + P - Command palette

Code Explanation

Select code and ask Codex to explain it:

1. Highlight code
2. Right-click → Codex: Explain
3. Get explanation from Claude 4.8

Refactoring

Ask Codex to refactor code:

1. Select code to refactor
2. Right-click → Codex: Refactor
3. Describe desired changes

Advanced Features

Custom Prompts

Create custom prompts in Codex settings:

{
  "customPrompts": [
    {
      "name": "Document Function",
      "prompt": "Add comprehensive documentation to this function including JSDoc comments, parameter descriptions, return value, and usage examples."
    },
    {
      "name": "Optimize Performance",
      "prompt": "Analyze this code for performance issues and suggest optimizations."
    }
  ]
}

Model Parameters

Fine-tune Claude's behavior:

{
  "api": {
    "baseUrl": "https://ai.xiaoyuan.co.uk/v1",
    "apiKey": "your-api-key-here",
    "model": "claude-opus-4-8",
    "parameters": {
      "temperature": 0.7,
      "max_tokens": 4096,
      "top_p": 0.9
    }
  }
}

Context Management

Configure how much code context Codex sends:

{
  "context": {
    "includeOpenFiles": true,
    "includeGitDiff": true,
    "maxTokens": 100000
  }
}

Verification

Test your setup:

1. Open Codex
2. Start a new chat
3. Ask: "What model are you?"
4. Claude should identify as Opus 4.8

Or run a code generation task to verify it's working.

Troubleshooting

Connection Failed

If Codex can't connect to XiaoYuan AI:

1. Check the base URL: https://ai.xiaoyuan.co.uk/v1
2. Verify your API key
3. Test with curl:

curl https://ai.xiaoyuan.co.uk/v1/models \
  -H "Authorization: Bearer your-api-key-here"

Slow Response Times

If responses are slow:

1. Check your internet connection
2. Reduce context size in settings
3. Lower max_tokens parameter
4. Check API status at https://ai.xiaoyuan.co.uk

Authentication Errors

Error: Invalid API key

Solutions:

• Verify API key is correct
• Regenerate API key at https://ai.xiaoyuan.co.uk
• Check for extra spaces in configuration
• Ensure key hasn't expired

Model Not Available

Error: Model not found

Ensure model name is exactly:

• claude-opus-4-8 (recommended)
• claude-opus-4.8
• claude-opus-4-latest

Tips & Best Practices

1. Use Specific Prompts

Instead of:

"Fix this"

Try:

"Refactor this function to improve readability, add error handling, and follow Python PEP 8 style guidelines"

2. Provide Context

Give Claude context about your project:

• Select relevant code files
• Describe the project structure
• Mention frameworks and libraries used

3. Iterative Refinement

Work iteratively:

1. Generate initial code
2. Review and test
3. Ask for specific improvements
4. Repeat until satisfied

4. Leverage Chat History

Claude maintains conversation context:

• Reference previous exchanges
• Build on earlier suggestions
• Ask follow-up questions

Integration with IDEs

VS Code

Install Codex extension for VS Code:

code --install-extension codex.vscode-codex

Configure in VS Code settings.json:

{
  "codex.api.baseUrl": "https://ai.xiaoyuan.co.uk/v1",
  "codex.api.apiKey": "your-api-key-here",
  "codex.api.model": "claude-opus-4-8"
}

Neovim

Add to your Neovim config:

require('codex').setup({
  api = {
    base_url = 'https://ai.xiaoyuan.co.uk/v1',
    api_key = os.getenv('XIAOYUAN_API_KEY'),
    model = 'claude-opus-4-8'
  }
})

Support

Need help?

1. Check Getting Started
2. Review API Key guide
3. Visit https://ai.xiaoyuan.co.uk for support