Termexo

Product Introduction

1. Definition: Termexo is a local-first, Windows-native desktop application that functions as a unified workspace and session manager for AI-powered coding agents like Claude Code and Codex. It operates as a terminal multiplexer with intelligent agent-aware features.
2. Core Value Proposition: Termexo exists to eliminate the friction and context-switching overhead of managing multiple AI coding agent sessions. Its core value is providing a single, persistent, and recoverable window where developers can run, monitor, and control several agents simultaneously, dramatically improving workflow efficiency for AI-augmented coding.

Main Features

1. Customizable Terminal Grid Workspace: Termexo allows users to open multiple native PTY (Pseudo-Terminal) terminals within a single window and arrange them in a customizable grid layout (from 1x1 to 6x6). Each project or "workspace" remembers its specific layout, open terminal sessions, and even color schemes, enabling a tailored, multi-agent development environment.
2. Cross-Agent Session Awareness & Notification System: The application actively monitors the state of all connected AI agents (Claude Code, Codex CLI). It visually distinguishes between states like "running," "thinking," "waiting for approval," and "finished" within tabs and a global panel. Critically, it sends system-wide notifications (taskbar flashing, system alerts) when an agent is stuck awaiting input, ensuring no work is blocked by an unfocused window.
3. Native Session Resume & Search: Termexo does not create its own session files. Instead, it indexes the native session history files created by Claude Code and Codex. Users can search across all projects by session name, file path, Git branch, or model used. Resuming a session leverages the official CLI commands (claude --resume, codex resume), guaranteeing full context restoration without data alteration.
4. Dynamic Model Profile Routing: Users can define multiple model profiles (e.g., for Anthropic Claude, DeepSeek, MiniMax, GLM, or custom MCP-compatible endpoints). These profiles, which include endpoint configuration, are switchable per-terminal or for an entire workspace with two clicks. API keys are stored securely in Windows Credential Manager, not in plain-text config files, separating secrets from application settings.

Problems Solved

1. Pain Point: Context fragmentation and agent sprawl. Developers using multiple AI coding assistants end up with numerous disjointed terminal windows and CMD/PowerShell instances, making it difficult to track which agent is working on what, leading to lost sessions and inefficient screen real estate use.
2. Target Audience: The primary user personas are Windows-based software engineers, full-stack developers, and technical leads who regularly utilize CLI-based AI coding agents like Claude Code and Codex for daily development, code review, and debugging tasks. It is especially valuable for developers managing multiple projects or experimenting with different AI models.
3. Use Cases: Essential for scenarios involving parallel AI-assisted workstreams, such as using Claude Code for feature development in one terminal while using Codex for generating tests in another, all within the same codebase. It is also critical for resuming complex, multi-step AI coding sessions from a previous day without manual reconstruction of context.

Unique Advantages

1. Differentiation: Unlike generic terminal multiplexers (e.g., tmux on Windows via WSL), Termexo is purpose-built with deep integration for specific AI coding agents, offering native session resume, agent state detection, and model routing. Compared to simply opening multiple terminal windows, it provides structured workspace management, persistent state, and proactive notifications.
2. Key Innovation: Its "local-first, agent-aware" architecture. Termexo innovates by acting as a thin, intelligent management layer on top of existing, official CLIs. It enhances their functionality (with search, unified UI, notifications) without replacing or intercepting their core operations, ensuring compatibility and respecting the user's existing workflow and data integrity (read-only session access).

Frequently Asked Questions (FAQ)

1. Is Termexo a cloud service or does it send my code/data online? No, Termexo is a strictly local-first Windows application. Your project code, AI session histories, and workspace configurations are stored only on your local machine. It has no mandatory cloud sync or account system. Future remote-access features are planned to be opt-in and user-controlled.
2. How does Termexo handle my Anthropic or other API keys securely? Termexo uses Windows Credential Manager (Windows Vault) for secure API key storage. The application itself only stores a reference to the credential and can check for its existence but cannot retrieve or display the plain-text key, providing a more secure alternative to plain-text configuration files.
3. Can I use Termexo with AI agents other than Claude Code and Codex? Currently, Termexo provides deep, state-aware integration for Claude Code and Codex CLI. It also supports routing to other Claude-compatible models/endpoints (like DeepSeek, MiniMax) via its profile system and can host native PTY terminals for any shell command, but advanced features like state detection are optimized for the primary supported agents.
4. What happens to my existing Claude Code sessions when I start using Termexo? Nothing. Termexo reads your existing session files in their default locations. It does not modify, move, or delete them. You can search and resume these past sessions directly from within Termexo, and any new sessions you start will be saved to the same location, fully compatible with the standard claude CLI.
5. Do I need to create an account or pay for Termexo? No. Termexo requires no account, subscription, or login. It is distributed via npm and as a standalone Windows installer from GitHub. The application runs entirely on your local machine with no paywall or required central service.