The agent that evolves with you, works for you and does not stop its autonomous loops untill it achieves its goal.
Install · Features · Commands · Develop
One command, any platform:
# macOS, Linux, Termux
curl -fsSL https://raw.githubusercontent.com/Clioloop/Clioloop-agent/main/scripts/install.sh | bash# Windows PowerShell
iex (irm https://raw.githubusercontent.com/Clioloop/Clioloop-agent/main/scripts/install.ps1)Windows users can also download Clio-Setup.exe — a GUI installer that handles everything.
After install:
clio # Launch interactive chat
clio desktop # Launch the desktop app
clio dashboard # Launch the web dashboard
clio update # Update to the latest versionThe fastest way to get running: pick Omni Loop Portal in clio setup, approve the
device code in your browser, and your agent gets every frontier model (Claude, GPT,
Gemini, Grok, DeepSeek, …) through a single subscription — no API keys to manage.
Works across the CLI, TUI, desktop app, and web dashboard. Plans start free
(card verification only). The portal itself lives in portal/ and is
self-hostable: cd portal && npm run setup && npm run dev.
Agentic Fusion is Clioloop's multi-model reasoning mode for Pro plans and up.
Planner/advisor models propose routes, your current main model performs the visible
full-tool work, reviewer models critique the draft, and the run fuses into one final
answer. It is available anywhere you can chat with Clioloop: CLI, --tui, desktop,
Telegram/gateway, and the managed portal flow.
Fusion works best, and is safest for advisor/reviewer usage, in a fresh session. Start a new session before every Fusion run, then reset again after the final fused answer:
/fusion status # inspect current Fusion config
/fusion # choose advisors/reviewers interactively
/new # or /reset: clear stale context before the next Fusion run
If you keep using an old session after a Fusion run, advisors and reviewers may inherit stale context and waste managed usage. Clioloop also shows a reset reminder after each completed Fusion run, but the reset itself is intentionally manual.
Full-featured agent with tool calls, multi-turn conversations, context management, and support for OpenAI, Anthropic, Google Gemini, Groq, OpenRouter, and any OpenAI-compatible endpoint.
Clioloop doesn't just run — it learns. The agent detects repeated work patterns and generates reusable skills from them. Skills improve during use and can be shared via the Skills Hub.
Built for the command line. The TUI (terminal UI) provides a rich chat experience with markdown rendering, syntax-highlighted code blocks, inline images, and session management — all inside your terminal.
| Surface | Command | Description |
|---|---|---|
| Terminal TUI | clio |
Full-featured terminal chat UI |
| Desktop App | clio desktop |
Electron-based GUI with system tray |
| Web Dashboard | clio dashboard |
Browser-based management interface |
| Headless API | clio gateway |
REST/WebSocket API for integrations |
| Slack Bot | clio slack |
Chat with Clioloop from Slack |
| Telegram Bot | clio gateway |
Telegram integration via the gateway |
clio whatsapp |
WhatsApp messaging support | |
| VS Code | clio lsp |
LSP-powered editor integration |
Swap models and providers at runtime:
clio -m gpt-4o # OpenAI
clio -m claude-sonnet-4-20250514 # Anthropic
clio -m gemini-2.5-pro # Google
clio --provider openrouter # Any OpenRouter modelWeb browsing, file operations, code execution (Python/JS/bash), image generation, audio transcription, browser automation, Discord, email, calendar, Jira, and more — all configurable per session.
Long-term memory via vector databases, session persistence with checkpoint/restore, cron-based scheduled tasks, and kanban-style project management.
SOC2-aware audit logging, PII redaction, credential scanning, file access controls, and a doctor command that validates your entire setup.
| Command | Description |
|---|---|
clio |
Start an interactive chat session |
clio chat "message" |
One-shot query (non-interactive) |
clio acp "task" |
Agent Communication Protocol — autonomous task execution |
clio auth |
Manage API keys and provider credentials |
clio backup |
Backup your Clioloop configuration |
clio config |
View and edit configuration |
clio cron |
Schedule recurring agent tasks |
clio dashboard |
Launch the web dashboard |
clio desktop |
Launch the desktop GUI app |
clio doctor |
Diagnose installation and configuration issues |
clio gateway |
Start the REST API gateway |
clio gui |
Launch the Tauri-based installer |
clio kanban |
Kanban-style task management |
clio login |
Authenticate with providers |
clio mcp |
Model Context Protocol server |
clio memory |
Manage long-term memory stores |
clio model |
List and switch AI models |
clio plugins |
Manage plugins and integrations |
clio profile |
Switch between agent profiles/personas |
clio security |
Security audit and advisory checks |
clio setup |
First-run setup wizard |
clio skills |
Browse, install, and manage skills |
clio status |
Show system status and version info |
clio tools |
Configure available tool sets |
clio uninstall |
Remove Clioloop from your system |
clio update |
Update to the latest version |
clio version |
Print version information |
| Command | Description |
|---|---|
/fusion |
Open the Agentic Fusion picker |
/fusion status |
Show current advisors, reviewers, mode, and last run status |
/fusion off |
Disable Fusion for the current config |
/new or /reset |
Start a clean session before the next Fusion run |
# Clone and set up
git clone https://github.com/Clioloop/Clioloop-agent.git
cd clioloop-agent
# Python dependencies (uv recommended)
uv sync --all-extras --dev
# Node dependencies (for desktop, TUI, and web)
npm install
# Run from source
uv run clio --help
# Run tests
npm -w ui-tui test # Terminal UI tests
uv run pytest tests/ # Python test suite
# Build desktop app
npm -w apps/desktop run build
# Build Windows installer
npm -w apps/bootstrap-installer run tauri:buildclioloop-agent/
├── clio_cli/ # CLI frontend & Python agent core
├── apps/
│ ├── desktop/ # Electron desktop GUI
│ ├── bootstrap-installer/ # Tauri Windows installer
│ └── ... # Other app surfaces
├── ui-tui/ # Terminal UI (React Ink)
├── web/ # Web dashboard (React)
├── agent/ # Agent runtime & tool execution
├── gateway/ # REST/WebSocket API gateway
├── acp_registry/ # Agent Communication Protocol registry
├── tools/ # Built-in tool implementations
├── scripts/ # Install scripts for all platforms
├── docs/ # Documentation
└── tests/ # Test suite
MIT. See LICENSE.
Built by Omni Loop Labs