https://github.com/aiassistsecure/keystone-lite
A local-first desktop editor for AI-assisted debugging and surgical code edits.
https://github.com/aiassistsecure/keystone-lite
agentic agentic-workflow agentic-workflows ai ai-assist-secure code-debugger llm llm-coder vibe vibe-coding vibe-coding-assistant
Last synced: 2 months ago
JSON representation
A local-first desktop editor for AI-assisted debugging and surgical code edits.
- Host: GitHub
- URL: https://github.com/aiassistsecure/keystone-lite
- Owner: aiassistsecure
- License: other
- Created: 2026-01-17T22:22:39.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2026-01-18T03:46:01.000Z (2 months ago)
- Last Synced: 2026-01-18T14:59:04.893Z (2 months ago)
- Topics: agentic, agentic-workflow, agentic-workflows, ai, ai-assist-secure, code-debugger, llm, llm-coder, vibe, vibe-coding, vibe-coding-assistant
- Language: TypeScript
- Homepage: https://aiassist.net
- Size: 1.42 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# KeyStone Lite
**Your AI pair programmer, running locally.** Ship code faster with surgical AI edits that actually work.
KeyStone Lite is a desktop code editor with built-in AI that understands your codebase and makes precise, line-level changes. No more copy-pasting from ChatGPT. No more broken suggestions. Just ask, review, apply.
## Interface Overview
## Why KeyStone?
### For Developers
- **Actually understands your files** - Your active file is automatically included with line numbers, so the AI knows exactly where to make changes
- **Surgical precision** - No more "here's the whole file rewritten." KeyStone uses INSERT, REPLACE, DELETE operations on specific lines
- **11+ LLM providers** - Groq, OpenAI, Anthropic, Gemini, Mistral, Cohere, DeepSeek, or your own local Ollama instance
- **Two modes for two mindsets** - Debug mode lets you review before applying. Keystone mode auto-applies for rapid iteration
### For Founders & Teams
- **Ship MVPs faster** - Stop context-switching between AI chat and your editor. It's all in one place
- **Bring your own keys** - Use your existing API keys. Pay for what you use, no per-seat SaaS pricing
- **White-label ready** - Fork it, brand it, ship it as your own dev tool (see licensing)
## Quick Start
1. **Get an API key** from [aiassist.net](https://aiassist.net)
2. **Download** the latest release for your platform
3. **Open a project folder** and start coding with AI
```bash
# Or run from source
cd keystone-lite
npm install
npm run dev # Start Vite dev server
npm start # Launch Electron (in another terminal)
```
## Features at a Glance
| Feature | Description |
|---------|-------------|
| **Multi-file context** | Add multiple files to AI context with one click |
| **Surgical edits** | Line-precise INSERT, REPLACE, DELETE operations |
| **Mode toggle** | Debug (review first) or Keystone (auto-apply) |
| **Project templates** | Static HTML/TailWindCSS, React, Express, FastAPI, Electron starters |
| **Monaco editor** | Full VS Code editing experience |
| **Custom endpoints** | Connect any OpenAI-compatible API |
| **Secure by default** | Files sandboxed to project, keys encrypted locally |
## How It Works
### 1. Open Your Project
Click "Open Folder" or drag a directory. Your file tree appears in the sidebar.
### 2. Chat With Your Code
Open a file - it's automatically added to the AI's context with line numbers. Ask questions or request changes.
### 3. Review & Apply
The AI responds with precise edit blocks:
```
<<>>
<<>>
const handleSubmit = async (data: FormData) => {
const response = await api.post('/submit', data);
return response.json();
};
<<>>
```
In Debug mode, click "Apply All" to apply. In Keystone mode, changes apply instantly.
## Supported Providers
| Provider | Models |
|----------|--------|
| **Groq** | Llama 3.3 70B, Mixtral, Gemma 2 |
| **OpenAI** | GPT-4o, GPT-4 Turbo, o1 |
| **Anthropic** | Claude 3.5 Sonnet, Claude 3 Opus |
| **Google** | Gemini 1.5 Pro, Gemini 2.0 Flash |
| **Mistral** | Mistral Large, Codestral |
| **xAI** | xAI Grok |
| **TogetherAI** | Open-source at scale |
| **OpenRouter** | Multi-provider access |
| **DeepSeek** | DeepSeek Coder, DeepSeek Chat |
| **Fireworks AI** | Fast open-source |
| **Perplexity** | Search-augmented AI |
| **PIN Network** | Decentralized inference |
## Building for Distribution
```bash
npm run dist # Build for current platform
npm run dist:all # Build for Mac, Windows, Linux
```
Installers are output to the `dist/` folder.
## For Contributors
### Architecture
```
keystone-lite/
├── src/
│ ├── main/ # Electron main process
│ │ ├── index.ts # IPC handlers, file ops
│ │ └── preload.ts # Context bridge
│ └── renderer/ # React frontend
│ ├── components/ # ChatPanel, EditorTabs, ModelSelector
│ ├── lib/ # surgical-edit.ts parser
│ └── pages/ # MainLayout
├── templates/ # Project starters
└── assets/ # Icons and images
```
### Key Files
| File | Purpose |
|------|---------|
| `ChatPanel.tsx` | AI chat, API calls, system prompts |
| `ModelSelector.tsx` | Provider/model selection |
| `surgical-edit.ts` | Edit parsing and application |
| `MainLayout.tsx` | App layout and state |
| `index.ts` (main) | File system operations |
## Security
- API keys encrypted at rest via electron-store
- All file operations sandboxed to opened project
- No telemetry, no tracking, no phone-home
- Context isolation enabled in Electron
## License
**Business Source License 1.1** - See [LICENSE](./LICENSE)
- Free for personal use, development, and evaluation
- Production use requires AiAS API keys or commercial license
- Forks for distribution require written authorization from Interchained LLC
- Converts to MIT License on January 1, 2030
For commercial licensing: dev@interchained.org
## Links
- **API Keys & Platform**: [aiassist.net](https://aiassist.net)
- **Issues & PRs**: [GitHub](https://github.com/aiassistsecure/keystone-lite)
- **Enterprise Licensing**: dev@interchained.org
---
**Built by [AiAssist Secure (AiAS) of Interchained LLC](https://aiassist.net)** - Powering the next generation of AI-assisted workflows.