Local AI Assistant
A Visual Studio Code extension that connects to a local LLM (Large Language Model) API endpoint and provides code completion, validation, and query-via-chat functionality. Perfect for developers using private or offline models running on their workstation or on the local network.
Requirements
- A running local LLM API endpoint such as LM Studio, Ollama, KoboldCpp, or Oobabooga
- Network access to said endpoint (localhost or LAN)
- VS Code version
1.90.0 or newer
Connecting the extension to your LLM
The extension must be connected to your LLM before you can use it. Follow the steps below to connect it.
Install an LLM service provider such as LM Studio if you haven't already
Start the servce and host your model on the network
Install the extension in your VS Code
Launch the extension from status bar button (bottom left) or by using the keyboard shortcut CTRL + ALT + SHIFT + C
From the main UI, set up the URL, API Type (OpenAPI-compatible vs Ollama), and the model you want to use.
(Optional) Add an API key or specify a model if your service requires them
Features List
- Code validation / Code completion in chat
- Code completion (in-line) - Coming soon
- Tooltip / Right-click menu with extension commands
- File context management (add single file, all open files, clear context, add current file)
- Chat / Extension Settings UI
Validate Code Block
Checks and validates selected code (or entire file if nothing is selected).
- Windows:
Ctrl + Shift + Alt + V
- macOS:
Cmd + Shift + Alt + V
Auto-Complete Code in Chat
Autocompletes the current line or selection in chat using your configured LLM.
- Windows:
Ctrl + Shift + Alt + Enter
- macOS:
Cmd + Shift + Alt + Enter
Auto-Complete Code In-line
Open Chat Panel
Launches an interactive chat interface for general-purpose LLM interaction.
- Windows:
Ctrl + Shift + Alt + C
- macOS:
Cmd + Shift + Alt + C
Open Settings Panel
Quickly access and configure settings like endpoint, model, and behavior.
- Trigger via command palette:
Local AI Assistant – Open Settings
Editor menu with extension commands
Right-click to quickly access the extension's commands (useful if you want to disable the tooltip)
Shows contextual tooltips when idle, including quick links and actions.
API Key Support
Store an API key securely for authenticated LLM endpoints.
NOTE: This setting VS Code secure secrets storage API and does not store the key in settings.json
- Windows:
Ctrl + Shift + Alt + K
- macOS:
Cmd + Shift + Alt + K
Select LLM Model
Pick a model from your configured service if required (e.g. Qwen, Mistral, etc).
- Windows:
Ctrl + Shift + Alt + M
- macOS:
Cmd + Shift + Alt + M
Extension Settings
Accessible via:
- The gear icon in the chat panel
- Preferences → Settings → Extensions → Local AI Assistant
Configuration Options
API Settings
| Setting |
Description |
localAIAssistant.apiLLM.config.apiType |
Select LLM type (OpenAI, Ollama) |
localAIAssistant.apiLLM.apiURL.endpoint |
LLM API base URL (e.g. http://localhost:1234/v1) |
localAIAssistant.apiLLM.config.apiAuthRequired |
If enabled, prompts user to enter a secure API key |
localAIAssistant.apiLLM.config.model |
Default model for completions and chat (e.g. qwen2.5-coder-7b-instruct) |
Context Settings
| Setting |
Description |
localAIAssistant.context.contextSize |
AI model context limit (in tokens) |
localAIAssistant.context.includeFileContext |
Whether to send active editor text during code/chat requests |
UI Settings
| Setting |
Description |
localAIAssistant.tooltop.enableExtensionTooltip |
Enables or disables the tooltip in the main code editor |
localAIAssistant.tooltip.idleTooltipDelay |
Delay before showing idle tooltips (in milliseconds) |
Upcoming features
- Inline completion
- Status indicator for LLM service
- Token limit/character warnings
- Smart usage/embedding of current file in context
- SSL Validation
Known Issues
- Ollama support is limited
- API Key is optional and usage applies to OpenAI-based back-ends only
- Temporary freeze possible if endpoint is unreachable during stream
- Auto-scroll can be wonky
Author & Repository
Author: officedrone
GitHub: github.com/officedrone/local-ai-assistant
Issues: Report bugs or request features
License
GNU GPL v3
3rd party markdown used in chat
markdown-it.min.js
