Noureddine RAMDI / Clawd on Desk: an ambient desktop pet for real-time AI agent observability

Created Sat, 23 May 2026 20:41:14 +0000 Modified Sat, 23 May 2026 20:41:27 +0000

rullerzhou-afk/clawd-on-desk

Clawd on Desk flips the usual AI agent observability script by turning it into an ambient desktop pet — a pixel-art character that visually reflects what your AI coding assistants are doing in real time. Instead of switching back and forth between terminals, logs, and dashboards, you get a glanceable, interactive presence on your desktop that signals agent states and permission requests.

What Clawd on Desk does and its architecture

At its core, Clawd on Desk is an Electron app designed to integrate with multiple AI coding agents and tools, including Claude Code, Codex CLI, Cursor Agent, and more than nine others. It hooks into these agents through plugins and agent hooks, syncing their states and activities into a unified interface.

The app visualizes agent states using 12 distinct pixel-art animations, providing an ambient, glanceable UI rather than a traditional log or text-based interface. When an AI tool requests permissions — for example, to access a resource or execute an action — Clawd surfaces these as floating bubble cards. This design lets users approve or deny permissions inline without switching context back to a terminal or CLI.

Clawd also tracks multiple concurrent AI agent sessions, offering a dashboard and a compact heads-up display (HUD) for quick monitoring. The interface supports drag-and-drop repositioning from any state and includes an eye-tracking feature to subtly adjust the pet’s gaze toward your cursor or areas of interest.

Under the hood, the app is built on Electron, leveraging web technologies for cross-platform compatibility on Windows, macOS, and Linux. It supports custom themes and animation imports (notably Codex Pet animations), i18n with five languages, and system integrations such as system tray controls, Do Not Disturb mode, and auto-start hooks tied to Claude Code.

Technical strengths and design tradeoffs

What sets Clawd on Desk apart is its ambient UX approach to AI agent observability. Instead of passive log-watching or terminal-based permission prompts, it creates a consistent, interactive desktop presence that shows agent states visually and handles permissions inline.

This approach reduces the cognitive friction developers face when juggling multiple AI tools. Seeing an animated pet that reacts to agents finishing tasks or requesting permissions lets you “walk away” physically without losing situational awareness. The permission bubbles are a neat UX solution to the common annoyance of context-switching for approvals.

The multi-agent session tracking with a dashboard and HUD is also a thoughtful feature. It recognizes that developers often deal with parallel AI workflows and need a consolidated view.

From a code and architecture perspective, the Electron base allows cross-platform support but comes with tradeoffs. Electron apps tend to have larger memory and CPU footprints compared to native apps, which could be a downside if you’re sensitive to resource usage. However, the flexibility in UI design and web tech familiarity likely outweigh this for the target audience.

The pixel-art animations and eye tracking show attention to detail and polish, but they also add complexity. Accessibility could be a limitation here — users relying on screen readers or non-visual cues might not get the same benefit. Also, while the ambient pet is useful for quick glances, it’s not a replacement for deep debugging tools or detailed logs.

The app supports custom themes and animation imports, which opens up extensibility for users wanting to personalize their experience or add new AI agents. The system tray integration and Do Not Disturb mode further indicate a mature focus on fitting into real workflows.

Quick start

For users looking to try Clawd on Desk without diving into source code or builds, the project provides prebuilt installers for major platforms via GitHub Releases:

  • Windows: Clawd-on-Desk-Setup--x64.exe or Clawd-on-Desk-Setup--arm64.exe
  • macOS: .dmg installer
  • Linux: .AppImage or .deb packages

Once installed, launching Clawd automatically syncs supported agent hooks and plugins on startup. This seamless syncing means you don’t need additional configuration to connect to your AI agents.

If you’re interested in contributing, testing unreleased features, or debugging, you can run the app from source. Note that this requires a full npm install and Electron build tooling, which can result in a large node_modules directory.

Here’s the exact command from the README for developers:

# Install dependencies
npm install

Verdict

Clawd on Desk is a niche but clever project that reimagines AI agent observability as an ambient desktop experience rather than a terminal or dashboard chore. It’s particularly relevant for developers who interact with multiple AI coding assistants and want a less intrusive way to monitor progress and approve permissions.

The Electron stack and pixel-art UI provide cross-platform reach and a playful vibe, but at the cost of heavier resource usage and potential accessibility tradeoffs. The ambient pet can’t replace detailed logging or debugging but complements those workflows by reducing context switches and keeping you visually connected to what your AI agents are doing.

If you’re already deep in the AI-assisted development space and juggling multiple agents like Claude Code and Codex CLI, this tool offers a unique layer of observability and control that’s worth exploring. For general AI or coding use, the overhead might not justify itself, but for power users, Clawd on Desk introduces a fresh UX angle that’s rare in this space.

Overall, it’s a solid open source effort with attention to detail and practical quick start options, though it’s not a universal solution. Worth a look if you want your AI coding helpers to be a pixel-art companion rather than just another CLI process.


→ GitHub Repo: rullerzhou-afk/clawd-on-desk ⭐ 2,807 · JavaScript