TealClaw
BYOK voice dashboard. Zero server. Total privacy.
Quick Setup
Paste any API key directly in the chat. TealClaw auto-detects the key type by its prefix and saves it. Type /setup for the guided wizard.
| Prefix | Detected As |
|---|---|
gsk_ | Groq (Whisper + AI) |
sk-or-v1- | OpenRouter AI |
sk-ant- | Anthropic AI |
sk_ + hex | ElevenLabs TTS |
digits:alpha | Telegram Bot Token |
http://...:port | Gateway URL |
| 32+ hex chars | Gateway Token |
https://.../images | Image Gen Endpoint |
For AI Agents
You can configure TealClaw instantly using a config URL. The #config= hash fragment never leaves the browser (URL fragments are not sent in HTTP requests).
Option 1: Config URL
Build a JSON object with the keys, base64-encode it, and construct:
https://tealclaw.com/#config=BASE64_ENCODED_JSON
Keys auto-populate into localStorage, the hash is stripped.
Option 2: Browser Console
localStorage.setItem('tc-keys', JSON.stringify({
aiProvider: 'openrouter',
aiKey: 'sk-or-v1-...',
whisperKey: 'gsk_...',
ttsKey: 'sk_...',
ttsVoice: '4YYIPFl9wE5c4L2eu2Gb',
mode: 'agent'
}));
localStorage.setItem('tc-agents', JSON.stringify([
{id:'home', name:'Home', url:'http://machine.ts.net:18789', token:'tok', active:true}
]));
location.reload();
Architecture
Browser -> AI Provider (OpenRouter/Groq/Anthropic) -> Browser
Browser -> Groq Whisper (transcription) -> Browser
Browser -> ElevenLabs (TTS) -> Browser
Browser -> OpenClaw Gateway (via Tailscale) -> Browser
Browser -> Telegram Bot API -> Telegram Chat
No backend. No database. No tracking. Static files on Cloudflare Pages. Open DevTools > Network to verify.
Image Generation
Type /imagine your prompt here in the chat to generate images. Paste an image gen endpoint URL in the chat to configure it.
Supported APIs
Any OpenAI-compatible image generation API works:
| Provider | Endpoint URL | Where to Get Key |
|---|---|---|
| OpenAI DALL-E | https://api.openai.com/v1/images/generations | platform.openai.com |
| Together AI | https://api.together.xyz/v1/images/generations | api.together.xyz |
| FAL.ai | https://fal.run/fal-ai/flux/dev | fal.ai/dashboard |
| Fireworks | https://api.fireworks.ai/inference/v1/images/generations | fireworks.ai |
Paste the endpoint URL in the chat to configure it. The API key falls back to your AI key if not set separately.
Usage
/imagine a teal cat with glowing eyes in cyberpunk style
/imagine watercolor painting of a sunset over mountains
/imagine minimalist logo design, flat vector art
Features
- Chat -- multi-turn with persistent history
- Voice In -- hold mic button, Groq Whisper transcription
- Voice Out -- ElevenLabs TTS, auto-plays on desktop; tap "Enable Audio" on mobile
- Image Gen -- type
/imagineto generate images via any compatible API - Camera -- capture photos, flip camera, flashlight
- Vision -- drag-and-drop, paste, attach images
- Multi-Agent -- add multiple gateways, switch in chat
- Telegram -- forward conversations to your bot
- Dark/Light -- theme toggle, responsive design
- PWA -- installable, works offline
Security Verification
- Open DevTools > Network tab
- Send a test message
- Confirm all requests go to AI provider, Groq, ElevenLabs, or your Tailscale gateway
- Confirm zero requests go to tealclaw.com after page load
Buy Me a Coffee -- TealClaw is free and open. Tips help keep development going.