OpenCompanion

Product Introduction

1. Definition: OpenCompanion is a local-first, open-source desktop application built with Tauri (Rust + WebView) and Svelte. It functions as a centralized management console and remote control interface for AI-powered command-line interface (CLI) coding assistants.
2. Core Value Proposition: It exists to solve the operational overhead and context-switching penalty for developers running multiple, long-lived AI CLI sessions (like Claude Code, Codex CLI, OpenCode) concurrently. Its primary value is enabling multi-session AI CLI management, providing centralized monitoring, cross-device remote approval, and session automation without relying on cloud services or compromising privacy.

Main Features

1. Unified Session Dashboard: The application provides a single-pane-of-glass view for all active AI CLI sessions across different project directories. It detects when a session is blocked waiting for user input (e.g., Claude Code's permission prompts) and provides visual and system-level notifications. This is implemented by spawning and managing CLI subprocesses, parsing their standard output streams in real-time, and applying pattern matching for known prompt formats.
2. Local Network Phone Companion: A key innovation is the integrated web server that allows secure, local-only remote access. Users can pair a smartphone or secondary device on the same Wi-Fi network via QR code or manual code entry. This companion interface, served over plain HTTP on the local network, mirrors the desktop dashboard, enabling users to monitor sessions, approve/deny requests, start new tasks, and send messages from their phone. For remote access, it recommends using a VPN like Tailscale.
3. Chat-Based Planning & Automation: Beyond simple monitoring, OpenCompanion includes a Chat interface where users can describe a coding task in natural language. A built-in planner parses this description and breaks it down into actionable session cards. Furthermore, an Automations feature allows scheduling CLI sessions using cron-like expressions or simple daily timers, enabling automated, headless execution of AI coding tasks.
4. Integrated Development Environment (IDE) Adjacent Tools: The app includes practical utilities for each session: a file browser for the project directory, a Git diff viewer for uncommitted changes, branch switching, and a multi-pane terminal emulator (supporting up to four split shells) that opens directly in the session's working directory, reducing the need to switch to a separate terminal or IDE.

Problems Solved

1. Pain Point: Context Switching and Lost Productivity. Developers running AI CLI agents in multiple terminals must manually check each window to see if a session is paused for approval, leading to significant downtime and disrupted workflow focus.
2. Target Audience: The primary user personas are Software Engineers, DevOps Engineers, and Technical Leads who actively use AI coding assistants like Claude Code for automation, refactoring, or code generation across several repositories or microservices simultaneously. It is also valuable for researchers or data scientists using these tools for script generation and analysis.
3. Use Cases:
   • Parallel Project Refactoring: Running Claude Code in "Plan" mode across 5 different repositories to suggest architecture improvements, with all statuses and approval requests centralized.
   • Scheduled Code Maintenance: Using Automations to run Codex CLI every night in a headless mode to apply security linting fixes across a monorepo.
   • Remote Development Oversight: Starting a long-running OpenCode session on a desktop machine and going to a meeting, while retaining the ability to monitor progress and approve critical actions from a phone.
   • Batch Task Execution: Using the Chat planner to describe a multi-step workflow (e.g., "Update dependencies, run tests, generate migration") and having OpenCompanion create and manage the discrete CLI sessions.

Unique Advantages

1. Differentiation: Unlike running individual terminal tabs or using generic process managers (like tmux or screen), OpenCompanion is purpose-built for the interaction patterns of modern AI CLIs. It understands session states (running, waiting) and specific prompt formats. Compared to potential cloud-based manager services, its strict local-first, zero-telemetry architecture ensures all code, AI interactions, and project data remain on the user's machine and local network.
2. Key Innovation: The seamless, account-less phone pairing system is a significant technical and UX innovation. It removes the complexity of account creation, cloud relays, and complex networking setup (like port forwarding) for a common use case. The combination of fine-grained permission modes (Ask, Plan, Auto, Bypass) per session with cross-device control creates a uniquely flexible governance model for AI-assisted coding.

Frequently Asked Questions (FAQ)

1. Is OpenCompanion a cloud service or does it send my code/data online? No. OpenCompanion is a local desktop application with a local network web interface. It requires no account, has no cloud server component, and collects zero telemetry. Your AI CLI sessions communicate directly with their respective providers (Anthropic, OpenAI, etc.) as they normally would; OpenCompanion only manages the process and facilitates your local input.
2. How do I use OpenCompanion with my phone securely outside my home network? The phone companion is designed for local Wi-Fi. For secure remote access, the documentation recommends using a VPN solution like Tailscale or ZeroTier to create a secure virtual network between your desktop and phone, treating it as a "local" connection. Do not expose the OpenCompanion HTTP port directly to the public internet.
3. Does OpenCompanion install or modify my AI CLI tools (Claude Code, Codex CLI, etc.)? No. OpenCompanion only detects CLIs already available on your system's PATH. It does not install, update, or modify them. You must install and authenticate with each CLI (e.g., claude auth login) independently before using them within OpenCompanion.
4. What operating systems and AI CLIs are supported? OpenCompanion provides installers for Windows (MSI/EXE), Linux (DEB, RPM, AppImage), and macOS (DMG for Apple Silicon/Intel). It has confirmed support for managing sessions from Claude Code, Codex CLI, OpenCode, CCS, Pi, and omp, with varying levels of feature support (e.g., remote approval is fully functional for Claude Code and CCS).
5. Can I contribute to or modify OpenCompanion since it's open source? Yes. OpenCompanion is released under the MIT License, allowing free use, modification, and distribution. The repository includes comprehensive contributing guidelines, a project layout document, and uses a modern stack (Rust, Tauri 2, Svelte, Bun, Vitest) welcoming community bug reports, feature ideas, and pull requests.