Built on WebClaw

OpenCami 🦎

A feature-rich desktop web client for OpenClaw

Multi-session chat with real-time streaming, voice I/O, workspace tools (files, memory, agents, skills, cron), AI personas, smart titles, browser notifications — 35+ features for power users.

Self-hosting tip: after OpenClaw updates, if OpenCami can’t connect or you see scope/auth errors, restart the Gateway first. If it’s still broken, check the OpenCami GitHub for the current workaround.

View on GitHub Get OpenClaw

# One-line install

$ curl -fsSL https://opencami.xyz/install.sh | bash

# Or via npm

$ npm install -g opencami && opencami

⚡

Real-Time

Token-by-token streaming. See responses as they're written.

🧠

Smart

AI titles, follow-ups, context meter. It thinks ahead.

🎨

Yours to Customize

Models, personas, themes, text size. Make it fit you.

📂

Full Workspace

File explorer, search, export, keyboard shortcuts. Desktop-first.

💬 Communication

3 ▼

⚡

Real-Time Streaming

Token-by-token via persistent WebSocket → SSE. Tool call indicators, smart polling fallback.

🔊

Voice Playback (TTS)

Multi-provider: ElevenLabs → OpenAI TTS → free Edge TTS fallback. Listen to any response.

🎤

Voice Input (STT)

Record voice messages instead of typing. ElevenLabs Scribe v2 → OpenAI Whisper → Browser (free). Auto language detection for 90+ languages with recording UI, timer & pulse animation.

📁 Organization

8 ▼

📁

Session Folders

Auto-grouped: Chats, Sub-agents, Cron/Isolated, Other. Clean sidebar.

📌

Pin Sessions

Pin important conversations to the top. Always accessible.

🗑️

Bulk Delete

Select multiple sessions, delete at once with confirmation dialog.

🛡️

Protected Sessions

Main + channel-bound sessions auto-protected from accidental deletion.

🤖

Agent Manager (Beta)

Full agent CRUD from the sidebar — create, edit, delete agents visually. Shows model, workspace, sandbox mode, tools & channel bindings.

📋

Cron Jobs Panel (Beta)

Manage recurring automations from a dedicated panel. Beta/experimental and actively evolving.

🧠

Memory Viewer

Browse and edit MEMORY.md and daily memory files. Grouped sidebar: Daily notes (newest first) and Notes (alphabetical). Toggle in Settings.

🔔

Browser Notifications

Get notified when assistant replies while tab is in background. Debounced, permission-aware, click to jump to session.

🔍 Search & Navigation

3 ▼

🔍

Conversation Search

⌘F current chat, ⌘⇧F global deep search. Instant results with highlights.

⌨️

Keyboard Shortcuts

⌘N new chat, ⌘B sidebar, ⌘↵ send, ⌘⌫ delete. Full power-user nav.

📥

Export Conversations

Markdown, JSON, or plain text. Your data is yours.

🎨 Customization

8 ▼

🎨

Model Selector

Switch AI models per message from your gateway config.

🎭

Persona Picker

20 AI personalities. Integrates with ClawHub personas skill.

🦎

Chameleon Theme

Light/dark/system auto. Smooth transitions. Signature green.

🔤

Text Size

S / M / L / XL in Settings. Readable at any distance.

⚙️

TTS/STT Provider Selection

Choose your preferred voice provider in Settings. TTS: Auto / ElevenLabs / OpenAI / Edge TTS (free). STT: Auto / ElevenLabs / OpenAI / Browser (free).

🧠

Thinking Level Toggle

Brain icon in the composer to quickly switch reasoning depth for each prompt.

🔧

Multi-Provider LLM Settings

Configure OpenAI, OpenRouter, Ollama, or a custom endpoint from one unified settings panel.

🗂️

Workspace Settings

Unified toggles for Files, Memory, Agents, Skills & Cron. Show or hide each tool from the sidebar.

📂 Files

5 ▼

📂

File Explorer

Browse, upload, edit 30+ file types. Context menu, path jailing, breadcrumbs.

🖼️

Image Attachments

Drag & drop, paste, or click. Smart compression. Preview before send.

💻

Code Block Enhancements

Line numbers, filename headers, and a word-wrap toggle for cleaner code reading.

🔗

Smart File Links

Bare filenames become clickable links with quick directory navigation.

✏️

Open in Editor

Open file previews directly in the built-in editor with one click.

🧠 Smart Features

5 ▼

🏷️

Smart AI Titles

LLM-generated titles via gpt-4.1-nano. No more "New Chat #47".

💡

Smart Follow-ups

Context-aware suggestions. Instant heuristics → LLM refinements. One click.

📊

Context Window Meter

Token usage bar: green → yellow → red. Pulse warning at 95%+.

💬

Slash Commands

/help shows available commands. Quick access to OpenClaw features.

🔍

Search Sources Badge

Shows which search engines were used with favicons in an expandable badge. Supports Brave, Serper (Google), Tavily & Exa. Appears at end of response. Toggle in Settings.

📱 Mobile & PWA

5 ▼

👉

Swipe Gestures

Edge-swipe sidebar open/close. Dark backdrop, direction lock.

⚡

Performance

Lazy-loaded dialogs & routes. content-visibility. -16% bundle.

📱

Native Android APK (Beta)

Capacitor shell. App icons, splash screen, status bar integration.

🚀

PWA Support

Install as app, offline support, auto-update. Android & iOS.

🖥️

Desktop App (Beta)

Experimental Tauri v2 native wrapper for macOS/Windows/Linux with tray icon, native notifications, auto-start, custom titlebar, and multiple windows. Web app is still the primary focus.

⚡ 1.8.7

2026-03-07 ▼

Performance — Android PWA Overhaul

🚀 Streaming polling reduced 13× — 150ms → 2000ms. SSE handles real-time deltas; polling is now a safety net only
🎨 Render-blocking fonts removed — No more external Google Fonts import blocking first paint (200-800ms faster)
🧊 Frost theme mobile-optimized — Reduced backdrop blur on touch devices, preventing GPU-heavy repaints
💡 Deferred syntax highlighting — Code highlighting skipped during streaming, applied after response settles
🧩 Lazy-loaded UI components — Follow-ups, thinking indicators, search badges loaded on demand
📦 CSS containment — Added contain: content to messages + scroll compositing hints
🖼️ Image optimization — Memoized base64 URIs + low fetch priority for inline images

🔌 1.8.6

2026-03-07 ▼

Fixed

🔌 Duplicate WebSocket fix — Fixed Vite SSR issue causing doubled streaming text
🔌 Stale WebSocket cleanup — Proper cleanup of ghost connections on reconnect

🔐 1.8.5

2026-03-03 ▼

Fixed

🔐 Proper device pairing — correct scopes (operator.admin, operator.approvals, operator.pairing). No more allowInsecureAuth required.
🔐 deviceToken storage — after pairing, OpenCami stores a device token for future connections
🔐 Pending state detection — /api/ping now returns helpful approveCommand if device awaits gateway approval

🔒 1.8.4

2026-03-03 ▼

Security

🔒 Command injection fix — Skills API replaced execSync with execFile + argument arrays
🔒 SSRF / API key leak fix — LLM features API validates X-LLM-Base-URL against an allowlist
🔒 Path traversal consistency — File info endpoint uses shared jailed path helper
🔒 Private cache headers — User workspace files now served with private, no-store

🚀 1.6.0

2026-02-13 ▼

Added

🔧 Workspace Settings — Files, Memory, Agents, Skills & Cron toggles under dedicated section
🔔 Browser Notifications — Background tab alerts, debounced, permission-aware
🧠 Memory Viewer — Browse and edit MEMORY.md and daily files with grouped sidebar
📂 File Explorer toggle — controllable via Settings (default: on)

Fixed

Memory Viewer file paths correctly prefixed for API compatibility

⚡ 1.5.1

2026-02-12 ▼

⚡ Lazy-load Shiki — Dynamic imports for syntax highlighting
🔍 Promise.allSettled for global search — no more blocked results
🔒 LocalStorage API key warning in LLM Features
🧪 162 Vitest tests across 14 files
📚 5 new docs — API, Architecture, Contributing, Deployment, Features

🎨 1.5.0

2026-02-11 ▼

🖥️ Tauri Desktop App (Beta) — Native wrapper with tray, notifications, multi-window
🎨 Appearance Overhaul — Frost themes, accent colors, font family, density, bubble style
📦 Skills Browser (Beta) — Browse and discover ClawHub skills
📄 FAQ page
🔧 Fixed streaming race condition, bulk delete, search jump, Frost Dark CSS

🧩 1.4.0

2026-02-10 ▼

🧩 Code Block Enhancements — Line numbers, filename headers, word wrap
🤖 Cron Jobs Panel (Beta) — Visual manager with execution history
🔗 Smart File Links — Bare filenames clickable (35+ extensions)
📁 Directory Navigation, ✏️ Open in Editor
⚡ Search optimization, overflow hardening

🧠 1.3.2

2026-02-09 ▼

🧠 Thinking Level Toggle — Reasoning depth per message
🔌 Multi-Provider LLM — OpenAI, OpenRouter, Ollama, Custom
⚙️ Settings sidebar layout, 🎙️ Voice tab
🚀 Release automation (npm + GitHub Packages)

🎤 1.3.0

2026-02-08 ▼

🎤 Voice Input (STT) — ElevenLabs Scribe → OpenAI Whisper → Browser fallback
🔊 TTS/STT Provider Selection in Settings
🔍 Search Sources Badge with favicons
🤖 Agent Manager — Full CRUD from sidebar

📱 1.2.0

2026-02-07 ▼

👉 Swipe Gestures — Edge-swipe sidebar open/close
📱 Native Android APK (Capacitor)
⚡ Performance — Lazy dialogs/routes, -16% bundle
📊 Context Window Meter — Token usage visualization

🦎 1.0.0

2026-02-05 ▼

🚀 Initial release — Forked from WebClaw
⚡ Real-Time Streaming (WebSocket + SSE)
🔊 Voice Playback (TTS) — ElevenLabs/OpenAI/Edge
🎭 Persona Picker — 20 AI personalities
🎨 Model Selector, Chameleon Theme
📂 File Explorer, 🔍 Search, 📥 Export
📁 Session Folders, 📌 Pins, 🛡️ Protected Sessions
🏷️ Smart Titles, 💡 Follow-ups
📱 PWA, ⌨️ Keyboard Shortcuts, 💬 Slash Commands

⚛️

Modern Stack

Latest web technologies for performance and DX.

React 19TanStack StartTailwind 4TypeScriptVite

🔌

Gateway Architecture

WebSocket for real-time bidirectional communication with OpenClaw.

WebSocketSSE StreamingREST APICapacitor

🎯

Developer Friendly

Fork-ready, hot reload, upstream PR workflow. Easy to extend.

MIT LicenseHot ReloadDocumented

# Quick start

$ git clone https://github.com/robbyczgw-cla/opencami.git

$ cd opencami && npm install

$ npm run dev

🦎

"Started as a fork. Became a home."

OpenCami

🌐 Built with 💚 in the Clawiverse

GitHub OpenClaw Discord