clawtime

⚠Review·Scanned 2/19/2026

This skill sets up and runs the self-hosted ClawTime webchat, guiding installs, avatar creation, and service startup. It instructs executing shell commands (git clone, make, sudo tee, systemctl), performs network fetches (github.com, trycloudflare.com), and reads/writes ~/.clawtime/.env including GATEWAY_TOKEN.

from clawhub.ai·v88d9512·16.4 KB·0 installs

Scanned from 1.0.0 at 88d9512 · Transparency log ↗

$ vett add clawhub.ai/youngkent/clawtimeReview findings below

<p align="center"> <img src="public/apple-touch-icon.png" alt="ClawTime" width="120"> </p> <h1 align="center">ClawTime</h1> <p align="center"> A private, self-hosted web interface for your OpenClaw AI agent.<br> Voice conversations, animated avatars, end-to-end encryption. </p> <p align="center"> <img src="public/screenshot.jpg" alt="ClawTime Screenshot" width="700"> </p>

Overview

ClawTime is a webchat interface that connects directly to your local OpenClaw agent. No cloud services, no third-party servers — just you and your AI.

Unlike hosted chatbots where your conversations flow through corporate servers, ClawTime keeps everything on your own hardware. Your messages, history, and credentials never leave your machine.

Motivation

Most AI chat interfaces are either:

Cloud-hosted — your data goes to someone else's servers
Text-only — no voice, no visual feedback, feels robotic
Generic — same UI for everyone, no personality

ClawTime solves all three:

100% local — direct connection to your self-hosted agent
Multimodal — voice input, spoken responses, animated avatar
Customizable — your agent can modify its own interface

Key Features

🏠 Private by Design

Direct WebSocket connection to your OpenClaw gateway
All data stored locally (~/.clawtime/)
No analytics, no tracking, no data collection
E2E encryption (ECDH P-256 + AES-256-GCM)

🎤 Voice Conversations

Tap the avatar to start talking
Browser-based speech recognition
Server-side TTS via edge-tts (no API keys needed)
Barge-in support — interrupt anytime

🦞 Animated 3D Avatar

A voxel-style 3D character that reflects your agent's state:

Idle — gentle breathing, occasional blinks
Thinking — looking around, processing
Working — typing on laptop (extended tasks)
Talking — animated mouth, expressive movement
Listening — attentive posture (voice mode)
Happy / Error / Sleeping — emotional states

Fully customizable — create avatars that match your agent's personality.

🔐 Passwordless Authentication

WebAuthn passkeys (Face ID, fingerprint, security key)
No passwords to remember or leak
One-time setup token for first registration

🛠️ Agent-Modifiable

Your AI agent has full access to the codebase and can:

Customize the UI in real-time
Add new features
Fix issues without human intervention
Create custom avatars

Installation

Give this link to your AI agent:

https://github.com/youngkent/clawtime/blob/main/SKILL.md

The agent will handle the complete setup — install, configure, and deploy.

Requirements

OpenClaw gateway running locally
Node.js 18+
A tunnel for external access (Cloudflare, ngrok, or similar)

Architecture

┌─────────────┐     tunnel      ┌─────────────┐     WebSocket     ┌─────────────┐
│   Browser   │ ◄─────────────► │  ClawTime   │ ◄───────────────► │  OpenClaw   │
│  (mobile)   │     HTTPS       │   Server    │    localhost      │   Gateway   │
└─────────────┘                 └─────────────┘                   └─────────────┘
                                      │
                                      ▼
                                ~/.clawtime/
                                ├── .env (config)
                                ├── messages.json (history)
                                ├── credentials.json (passkeys)
                                └── avatars/ (custom)

License

Polyform Noncommercial 1.0.0 — see LICENSE

✅ Free for personal use, hobby projects, research, education
💼 Commercial use requires a paid license — tongkeyun@gmail.com

Author

Keyun Tong
GitHub: @youngkent
Email: tongkeyun@gmail.com