Skip to content

Latest commit

 

History

History
94 lines (80 loc) · 4.21 KB

CONFIGURATION.md

File metadata and controls

94 lines (80 loc) · 4.21 KB

CodeContinue Configuration Guide

Settings are stored in Packages/User/CodeContinue.sublime-settings. You can edit them directly via:

  • Preferences > Package Settings > CodeContinue > Settings
  • Or Command Palette (Ctrl+Shift+P / Cmd+Shift+P) >> Preferences: CodeContinue Settings

You can also re-run the interactive wizard anytime via CodeContinue: Configure or switch models on the fly with CodeContinue: Switch Model....


Settings Reference

Key Type Default Description
endpoint string "http://localhost:8000/v1/chat/completions" Base OpenAI-compatible completions endpoint URL (Required).
model string "" Model identifier to use for completions (Required).
api_key string "" API key for authentication (optional, leave blank for local servers).
chat_endpoint string "" Dual-Engine override: endpoint for Chat (inherits endpoint if blank).
chat_model string "" Dual-Engine override: model for Chat (inherits model if blank).
chat_api_key string "" Dual-Engine override: API key for Chat (inherits api_key if blank).
edit_endpoint string "" Dual-Engine override: endpoint for Edit Selection (inherits endpoint if blank).
edit_model string "" Dual-Engine override: model for Edit Selection (inherits model if blank).
edit_api_key string "" Dual-Engine override: API key for Edit Selection (inherits api_key if blank).
stream boolean true Stream chat responses in real-time using Server-Sent Events (SSE).
status_bar boolean true Display active model and status widget in Sublime Text's bottom status bar.
confirm_edits boolean true Require interactive confirmation and unified diff review before applying code edits.
max_context_lines integer 30 Number of surrounding lines of code sent as context to the model.
timeout_ms integer 30000 Request timeout in milliseconds (increase for slower or larger local models).
trigger_language array ["python", "cpp", "javascript"] Supported syntax languages/scopes (supports aliases like c++, py, ts, js, or ["*"]).
auto_suggest_on_enter boolean false When false, suggestions only trigger on-demand via shortcut. When true, triggers on Enter.
system_prompt string "" Custom system prompt for inline completions. Leave blank to use built-in default expert prompt.
debug boolean false Enable verbose logging to Sublime Text console (View > Show Console).

Provider Recipes

1. Local LM Studio

Start LM Studio, load your model, and start the local server (default port 1234):

{
    "endpoint": "http://localhost:1234/v1/chat/completions",
    "model": "qwen2.5-coder-3b-instruct"
}

2. Local Ollama

Run ollama serve (default port 11434):

{
    "endpoint": "http://localhost:11434/v1/chat/completions",
    "model": "qwen2.5-coder:3b"
}

3. OpenAI

Direct cloud connection to OpenAI:

{
    "endpoint": "https://api.openai.com/v1/chat/completions",
    "model": "gpt-4o",
    "api_key": "sk-your-openai-api-key"
}

4. Anthropic

Direct cloud connection using Anthropic's Messages API:

{
    "endpoint": "https://api.anthropic.com/v1/messages",
    "model": "claude-3-5-sonnet-20241022",
    "api_key": "sk-ant-your-anthropic-key"
}

5. Dual-Engine Setup (Fast Local Completions + Powerful Cloud Chat & Edit)

Use a low-latency local model for real-time autocomplete, while offloading code review, unit test generation, and deep refactoring to a flagship cloud model:

{
    // Fast local engine for inline completions
    "endpoint": "http://localhost:8000/v1/chat/completions",
    "model": "qwen2.5-coder:1.5b",

    // Flagship cloud engine for Chat about Selection
    "chat_endpoint": "https://api.anthropic.com/v1/messages",
    "chat_model": "claude-3-5-sonnet-20241022",
    "chat_api_key": "sk-ant-your-anthropic-key",

    // Flagship cloud engine for In-Place Code Editing
    "edit_endpoint": "https://api.openai.com/v1/chat/completions",
    "edit_model": "gpt-4o",
    "edit_api_key": "sk-your-openai-api-key"
}