← all repositories
ValueCell-ai/ClawX

A desktop shell for AI agents that refuses to make you touch a terminal

ClawX wraps the OpenClaw agent runtime in an Electron app so you can chat, schedule, and manage skills without ever writing YAML.

7.4k stars TypeScript AgentsLLMOps · Eval
ClawX
Velocity · 7d
+60
★ / day
Trend
steady
star history

What it does ClawX is an Electron + React desktop app that embeds the OpenClaw agent runtime and exposes it through a GUI. You get a chat interface, cron scheduling, multi-channel management (including a bundled WeChat plugin with QR login), and a skills panel—none of which require opening a terminal or editing config files by hand.

The interesting bit The app doesn’t just shell out to OpenClaw; it embeds the runtime directly, manages the gateway lifecycle, and even prunes stale bundled skills on startup to stay aligned with upstream. There’s also a “Dreams” page for memory review when Developer Mode is enabled, which is either genuinely useful or slightly unsettling depending on your stance on AI introspection.

Key highlights

  • Pre-bundled document-processing skills (pdf, xlsx, docx, pptx) auto-deploy on first launch
  • Per-agent provider/model overrides with @agent routing in the composer
  • Native keychain credential storage; OpenAI supports OAuth sign-in, not just API keys
  • Proxy settings with per-protocol overrides and automatic Telegram config sync
  • In-app QR flow for linking Tencent WeChat as a channel

Caveats

  • The README mentions an enterprise edition with “dedicated service support,” so the open-source build may be a loss leader
  • Some bundled skills (tavily-search) require API keys or will throw runtime errors
  • 4GB RAM minimum, 8GB recommended—not exactly lightweight for a chat wrapper

Verdict Worth a look if you want OpenClaw’s agent orchestration without the CLI tax. Skip it if you’re already comfortable running agents in a terminal or need a headless/server deployment.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.