TucDesk Platform
Enterprise-grade infrastructure management via intelligent terminal bridges, unified under a secure TUI and web dashboard. Powered by the TUC NLU engine and a vast marketplace of MCP AI connectors.
Overview
TucDesk is an end-to-end management layer that combines classical secure terminal access (SSH-like P2P tunnels) with modern AI-driven Natural Language Commands (NLC). It operates across three primary surfaces: Web Dashboard, TUI (Terminal User Interface), and Native Mobile (iOS/Android).
By decoupling the control plane from the data plane, TucDesk ensures that your actual terminal streams are end-to-end encrypted and never logged in plain text by the central server.
Core Features
- Zero-Trust P2P Networking: Direct agent-to-dashboard terminal streaming via WebRTC/QUIC tunnels.
- TUC Engine: A localized Natural Language Understanding model that translates text into CLI execution plans.
- Fallback LLMs: Seamless integration with 80+ AI models for resolving complex commands when local intent confidence is below 65%.
- Security Posture: Continuous automated audits with single-click remediation.
- Multi-Surface: Fully synchronized UI state across Web, TUI, iOS, and Android.
OS Installations
TucDesk agents are shipped as a single compiled binary with zero external dependencies.
For the full install matrix, package-manager options, mobile links, and direct artifacts, see the dedicated downloads page.
macOS
Use the installer script for a full setup, pin a specific release, or override the GitHub release namespace if you are shipping from a fork:
Linux (Ubuntu/Debian/RHEL)
The same shell installer works across Linux distributions. Direct archives and native packages are available when you need pinned rollouts or custom packaging.
Windows
Run the installer via PowerShell in an elevated prompt, or pin the release explicitly:
Desktop / Browser
The desktop control plane is also available as an installable PWA from the public web app. This is separate from the target-machine agent install.
iOS / Android
Mobile entry points live on the dedicated public pages: /ios and /android.
Package Managers
TucDesk can also be installed through package ecosystems if your team prefers native OS or language tooling.
Homebrew (macOS / Linux)
Windows package managers
Linux package managers
Language-native tools
MCP Connectors & Authentication
TucDesk integrates with 80+ AI providers via our Model Context Protocol (MCP) marketplace. This allows you to bring your own models (BYOM) to power the LLM-fallback capabilities of the TUC engine.
No API Keys Required in the Dashboard
We've overhauled the MCP integration so you do not need to copy and paste API secrets directly into TucDesk. Instead, TucDesk utilizes a secure SSO/OAuth flow directly with the providers.
How to Connect a Provider
- Navigate to the MCP Connectors tab in the Dashboard.
- Search for your preferred provider (e.g., Anthropic, OpenAI, DeepSeek).
- Click the Connect button on the provider's card.
- A secure window will open directing you to the provider's authentication portal.
- Log in to the provider (if required) to authorize TucDesk.
- The dashboard will automatically mark the provider as ✓ Connected and begin routing low-confidence intents to that model.
By using SSO authorization, your underlying API credentials remain safely managed by the upstream provider.
API Reference
The control plane API is available for automation and custom dashboards.
| Endpoint | Method | Description |
|---|---|---|
| /v1/auth/login | POST | Obtain an access and refresh token. |
| /v1/agents | GET | List all registered agents and their statuses. |
| /v1/security/run | POST | Trigger a synchronous security posture audit. |
| /v1/sessions | GET | List active terminal streaming sessions. |
Security & Compliance
TucDesk employs strict security constraints at every layer:
- Zero-Knowledge Tunnels: Control plane only negotiates the connection; it cannot decrypt the WebRTC payload.
- Operator Signatures: High-risk commands must be cryptographically signed by the dashboard client's Ed25519 key.
- Automated Audits: Running
/v1/security/runperforms checks on TLS configurations, admin passwords, and network bridging.