Skillget

DeskTalk

agent
Guvenlik Denetimi
Basarisiz
Health Uyari
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 6 GitHub stars
Code Basarisiz
  • rm -rf — Recursive force deletion command in packages/core/package.json
  • network request — Outbound network request in packages/core/package.json
  • process.env — Environment variable access in packages/core/playwright.config.ts
Permissions Gecti
  • Permissions — No dangerous permissions requested
Purpose: This project is an experimental, AI-native desktop environment that runs directly in your browser. It allows AI agents to control system applications, manipulate windows, and generate custom software ("LiveApps") on the fly.

Security Assessment: Overall risk is High. The tool fails a critical audit check for containing a recursive force deletion command (`rm -rf`) in its core package scripts, which poses a significant risk of accidental or malicious data loss. It also makes outbound network requests, likely to communicate with external AI providers. Environment variables are accessed via the testing configuration, which is normal but worth noting. Since the entire premise of the tool is allowing an AI to orchestrate and control a desktop environment, it inherently handles sensitive system operations.

Quality Assessment: The project is licensed under the permissive MIT license and was actively updated very recently (pushed to just 0 days ago). However, it suffers from extremely low community visibility, having only 6 stars on GitHub. This means the codebase has not undergone widespread peer review or security testing, making it firmly in the experimental or early-alpha stage.

Verdict: Not recommended for production or sensitive environments due to highly destructive script commands and the inherent risks of AI-controlled system orchestration.
SUMMARY

An AI-native desktop environment

README.md

NPM Version

DeskTalk

An AI-native desktop environment that runs in your browser.

What is this?

DeskTalk is an experiment in building an operating system designed for AI from the ground up. Instead of bolting AI onto traditional apps, every piece of the system is built to work with AI natively — while still being fully visible and usable by you.

tail-based-desktop

Goals

Fully AI-controlled system

DeskTalk is AI-native. Apps are designed for AI first. You describe what you want in natural language or voice, and the AI manipulates windows, invokes actions, edits content, and orchestrates your desktop on your behalf. Every app exposes actions the AI can call, making the entire system programmable through conversation.

Generative apps

You don't install most of your software in DeskTalk — you generate it. Tell the AI what you need and it creates a LiveApp: a self-contained, interactive application that runs directly on your desktop. LiveApps can be edited, persisted, and relaunched across sessions. Build dashboards, trackers, utilities, and visualizations — all through conversation.

Getting Started

Install DeskTalk from npm:

npm install -g @desktalk/core

Then start it:

desktalk start

That's it. Open your browser and you're in.

You can also customize the host and port:

desktalk start --host 0.0.0.0 --port 8080

Docker

Pull the published image from GitHub Container Registry:

docker pull ghcr.io/vincentdchan/desktalk:latest

Available tags are published from GitHub releases:

  • latest for stable releases
  • <version> such as 0.1.0
  • alpha for *-alpha.* releases
  • beta for *-beta.* releases
  • next for *-rc.* releases

Then run it with persistent volumes for DeskTalk data and config:

docker run -p 3000:3000 \
  -v desktalk-data:/home/node/.local/share/desktalk \
  -v desktalk-config:/home/node/.config/desktalk \
  ghcr.io/vincentdchan/desktalk:latest

Open http://localhost:3000 in your browser after the container starts.

The mounted volumes persist DeskTalk state across container restarts:

  • desktalk-data stores LiveApps, AI sessions, MiniApp data, user files, and embedded databases
  • desktalk-config stores DeskTalk configuration and AI provider credentials

If you want to use a different port on the host:

docker run -p 3000:3000 \
  -v desktalk-data:/home/node/.local/share/desktalk \
  -v desktalk-config:/home/node/.config/desktalk \
  ghcr.io/vincentdchan/desktalk:latest

If you want to build the image locally instead:

docker build -t desktalk .

LiveApps

LiveApps are the core user-facing concept. A user asks the AI for something — "build me a project tracker", "show me a chart of my disk usage" — and the AI generates a LiveApp: a self-contained HTML application that appears on the desktop and persists across sessions.

LiveApps can:

  • Execute shell commands and read system state through the DeskTalk bridge
  • Persist data with built-in KV and collection storage
  • Be edited in-place by the AI when you ask for changes
  • Use DeskTalk's full web component library for consistent, themed UI

Development

For development setup, build commands, testing, and release instructions, see docs/development.md.

Contributing

Contributions and suggestions are very much appreciated! Whether it's a bug report, a feature idea, or a pull request — all of it helps shape this project. Feel free to open an issue or submit a PR. We'd love to hear from you.

License

MIT

Yorumlar (0)

Sonuc bulunamadi