Skip to content

RunLLM integrates Large Language Models (LLMs) into PowerToys Run, enabling direct AI interaction from the search bar. It supports any service with an OpenAI API-style endpoint (e.g., Ollama, LMStudio).

License

Notifications You must be signed in to change notification settings

vuquan2005/RunLLM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

19 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RunLLM - PowerToys Run Plugin

Build Release License: MIT

🤖 Chat with Large Language Models directly from PowerToys Run

Features

  • Streaming Responses - Real-time text streaming from LLMs
  • Model Switching - Switch between models without leaving PowerToys Run
  • Thinking Mode - Toggle reasoning with /think (Qwen3) or enable_thinking (OpenAI)
  • Endpoint Switching - Change API endpoint on-the-fly with validation
  • API Key Support - Works with OpenAI, OpenRouter, and other authenticated APIs
  • Custom System Prompt - Define your own instructions

Requirements

Installation

From Release

  1. Download latest ZIP from Releases
  2. Extract to %LocalAppData%\Microsoft\PowerToys\PowerToys Run\Plugins\RunLLM
  3. Restart PowerToys

Folder structure after extraction:

Plugins/
└── RunLLM/
    ├── Community.PowerToys.Run.Plugin.RunLLM.dll
    ├── Community.PowerToys.Run.Plugin.RunLLM.deps.json
    ├── plugin.json
    └── Images/
        └── *.png

From Source

git clone https://github.com/vuquan2005/RunLLM.git
cd RunLLM
.\scripts\dev.ps1

Usage

Open PowerToys Run (Alt + Space) and type:

Command Action
runllm <question> Ask the LLM
runllm → Change Model Switch between available models
runllm → Thinking Mode Toggle thinking, select mode type
runllm → Change Endpoint Validate and set new API URL

Examples

runllm What is the capital of France?
runllm Explain async/await in JavaScript
runllm Write a Python function to reverse a string

Configuration

Go to PowerToys SettingsPowerToys RunPluginsRunLLM:

Setting Description Default
LLM URL API endpoint http://localhost:11434
Default Model Model name qwen/qwen3-4b
API Key Enable + enter key (disabled)
Enable Thinking Toggle thinking mode Off
System Prompt Custom instructions (empty)

Development

See DEVELOPMENT.md for build instructions.

.\scripts\dev.ps1      # Build + Deploy + Restart
.\scripts\build.ps1    # Build only
.\scripts\clean.ps1    # Clean artifacts

Architecture

src/
├── Main.cs            # Plugin entry point
├── QueryHandler.cs    # State machine, UI handlers
├── LLMClient.cs       # HTTP client, streaming
├── PluginSettings.cs  # Settings management
└── Constants.cs       # Default values

License

MIT License - see LICENSE.txt

About

RunLLM integrates Large Language Models (LLMs) into PowerToys Run, enabling direct AI interaction from the search bar. It supports any service with an OpenAI API-style endpoint (e.g., Ollama, LMStudio).

Topics

Resources

License

Stars

Watchers

Forks