Okan

Product Introduction

1. Definition: Okan is a specialized CLI-to-Browser bridge and notification system designed specifically for developers using Claude Code and similar AI agentic workflows. Technically categorized as a local-first development utility, it consists of a globally installed CLI package (okan-call), a persistent local background server, and a companion Chrome browser extension that facilitates bi-directional communication between the terminal and the web browser.

2. Core Value Proposition: Okan exists to eliminate the "context switching" friction and idle time inherent in long-running AI coding tasks. By surfacing terminal-based prompts and completion statuses directly within the browser UI, it allows developers to maintain productivity in their primary workspace (the browser) without missing critical interaction points or "Y/N" permission requests from Claude Code. Its primary mission is to minimize "Human-in-the-loop" latency in AI-assisted software development.

Main Features

1. Bi-Directional Permission Forwarding: Okan utilizes a WebSocket-based bridge to relay real-time terminal prompts to the browser. When Claude Code requests permission for file writes or command execution, Okan intercepts the request via a local hook and renders an interactive UI card in the Chrome browser. Users can approve or deny requests directly from the web interface, which then pipes the response back to the terminal process via the local server.

2. Multi-Tiered "Mom" Notification Modes: To ensure developer responsiveness, the tool features three distinct escalation levels for completion alerts:

• Gentle: Standard non-intrusive notification for casual monitoring.
• Classic: A balanced "Dinner is ready" alert (ご飯できたよ！) reflecting the product's persona.
• Mom: High-priority escalation where the notification card physically shakes on the screen after 5 seconds of inactivity, ensuring the developer returns to the terminal for task finalization.

3. Zero-Configuration Local Architecture: The system is engineered for immediate deployment. Upon running the global npm installation, the okan-hook automatically configures the necessary triggers. The local server handles discovery and connection between the browser extension and the terminal environment without requiring manual IP configuration, cloud accounts, or third-party API keys, ensuring a private, offline-capable workflow.

4. Dynamic UI Integration (Working Badge): While an AI task is active, a persistent "🍱 working..." status badge is injected into the bottom-right corner of the active browser tab. This low-profile visual indicator provides real-time state awareness of the background CLI process without occupying significant screen real estate.

Problems Solved

1. Context Switching Fatigue: Developers often lose focus when switching between a browser (for documentation or testing) and a terminal (for AI command execution). Okan solves this by bringing the terminal's interactive elements into the browser environment.

2. Terminal Idling: Complex refactoring tasks with Claude Code can take several minutes. Developers often switch to other tasks and fail to notice when the AI has paused to wait for permission. Okan eliminates this "dead time" by providing immediate visual and haptic (shaking) feedback the moment human input is required.

3. Target Audience: The tool is designed for software engineers, devops professionals, and AI researchers who utilize agentic CLI tools like Claude Code. It specifically targets those who manage multi-tasking workflows where the terminal is the execution engine but the browser is the primary viewing area.

4. Use Cases: Essential for large-scale codebase refactoring, automated test generation, and complex dependency updates where the AI agent frequently requires authorization to modify sensitive files or execute shell commands.

Unique Advantages

1. Differentiation: Unlike standard OS-level notifications (which are often silenced or ignored), Okan is interactive. It doesn't just tell you something happened; it allows you to take action (Approve/Deny) without leaving your current window. It bridges the gap between a notification and a remote terminal interface.

2. Key Innovation: The "Local-Only" security model is a significant advantage. By routing all data through a local WebSocket server (localhost) and keeping the logic on the user's machine, it avoids the security vulnerabilities associated with cloud-based notification relays, making it suitable for enterprise-level development on proprietary codebases.

3. Localized Persona: The product leverages a unique Japanese cultural theme ("Okan" or "Mom"), providing a more human-centric and engaging user experience compared to sterile, purely technical CLI utilities.

Frequently Asked Questions (FAQ)

1. Is Okan compatible with other AI tools like Cursor or Aider? Currently, Okan is optimized for Claude Code. However, the architecture is designed to be extensible, with support for Cursor and Aider listed on the development roadmap to provide a unified notification layer for all agentic CLI tools.

2. Does Okan send my code to an external server? No. Okan operates entirely on your local machine. The communication happens via a local server (localhost) between your terminal and your Chrome browser extension. No data is sent to the cloud, and no external accounts are required.

3. How do I change the notification intensity? You can toggle between Gentle, Classic, and Mom modes directly through the Chrome extension popup menu or via the command line using the okan config set mode [mode_name] command.

4. Does this tool work on browsers other than Chrome? Okan currently focuses on the Chrome Extension ecosystem. While it may work on Chromium-based browsers (like Edge or Brave) by loading the unpacked extension, official support is targeted at Google Chrome.