OmniTerminal

Inspiration

Managing modern multi-cloud infrastructure, remote Linux/Windows servers, and various database engines often requires jumping between multiple fragmented desktop tools. Existing solutions lack deep integration with AI to automate complex workflows safely. We built OmniTerminal to consolidate SSH, SFTP, RDP, and Database management (MySQL, PostgreSQL, Redis, MongoDB) into a single encrypted desktop application, powered by Google Gemini AI as an intelligent DevOps and Database assistant.

What it does

OmniTerminal is a high-performance, cross-platform desktop application (macOS, Windows, Linux) designed for server administration, database management, and automated DevOps playbooks:

• Google Gemini AI Assistant: Generates multi-step Ansible/Bash playbooks from natural language, analyzes live server telemetry (OS, CPU, RAM, disk usage, active services), and offers one-click execution with automated safety checks and undo/rollback stacks.
• Live Context Capture: Instant 1-click import of recent terminal outputs, active SQL queries, and error stack traces directly into Gemini for fast debugging and log anomaly detection.
• Multi-Protocol & Native DB Hub: Embedded high-performance SSH terminal (xterm.js), SFTP file manager with progress tracking, RDP auto-credential injection, and native query consoles for MySQL, PostgreSQL, Redis, and MongoDB.
• Enterprise Security & Passphrase Vault: AES-256-GCM encrypted local vault, KeePass-style keyfile protection, HashiCorp Vault REST API integration, built-in TOTP authenticator, and tamper-evident audit logging.

How we built it

• Frontend & Desktop App: Built using Electron, React, TypeScript, Vite, and Tailwind CSS.
• AI Integration: Integrated Google Gemini API via @google/genai to parse server context, answer queries, generate structured playbooks, and optimize system shell commands.
• Core Protocols: High-throughput streaming via Node.js native bindings, xterm.js for terminal UI, SSH2 for tunneling and bastion hosts, and native drivers for database engines.

Challenges we faced

• Streaming & Real-Time Context: Passing live server telemetry and large terminal log streams to Gemini without exceeding prompt sizes or degrading desktop app performance.
• Security & Isolation: Executing AI-generated shell and SQL scripts safely required designing pre-check dry-runs, high-risk command detection, and approval workflows before running commands against production nodes.

Accomplishments that we're proud of

• Delivering a fully functional, enterprise-grade desktop app with seamless Gemini AI integration.
• Enabling agentless telemetry capture where Gemini understands the exact environment context (OS, CPU cores, active services) in real time without needing background server daemons.
• End-to-end local vault encryption combined with HashiCorp Vault integration.

What we learned

• Effective prompt engineering techniques for structured JSON playbook generation and error stack trace resolution using Gemini models.
• Optimizing bidirectional streaming between Electron native processes and AI API endpoints for minimal latency.

What's next for OmniTerminal

• Multi-modal AI support for analyzing visual architecture diagrams and automated schema migrations.
• Gemini-powered autonomous root-cause analysis for dynamic multi-server log monitoring.

Built With

• electron
• gemini-api
• hashicorp-vault
• mongodb
• mysql
• node.js
• postgresql
• react
• redis
• ssh
• tailwind-css
• typescript