Agent Pets

Product Introduction

1. Definition: Agent Pets is a locally-hosted Chrome extension and Node.js-based helper application designed for developers. It functions as a unified notification and interaction layer for AI-powered coding agents like Claude Code, OpenCode, and omp.
2. Core Value Proposition: It exists to solve the context-switching problem inherent in using terminal-based AI coding assistants. By turning each active agent session into a persistent, animated "pet" in the browser, it allows developers to monitor status, approve permission requests, and send replies from any Chrome tab without interrupting their primary workflow on other windows or applications.

Main Features

1. Cross-Tab Agent Notification & Control: The core technical feature is a WebSocket-based communication bridge between a local helper daemon (127.0.0.1) and the Chrome extension. The helper integrates via official hooks (~/.claude/settings.json) and plugins (~/.config/opencode/plugins/) to intercept agent events. When an agent requires approval (e.g., for a bash command or file edit), the helper sends a message to the extension, causing the corresponding session's pet to animate and display a permission prompt on every active browser page.
2. Visual Session Management Dashboard: Each live session is represented by a unique, named pet (e.g., MangoCat, BubblesAxolotl) with a customizable avatar ("Your Cat"). A status capsule above each pet displays real-time agent states: "Thinking," "Your Turn," "Allow Bash?," or "Done." A badge on the extension icon aggregates and displays the total count of sessions awaiting user input, providing immediate visual workload awareness.
3. Local-First, Privacy-Centric Architecture: The system is engineered for complete data locality. All communication occurs between processes on the user's machine (localhost). The AI agent's code, prompts, and project context never leave the local environment. The only external network calls are for initial license validation (via Dodo Payments) and periodic update checks against the npm registry, making it suitable for sensitive or proprietary codebases.

Problems Solved

1. Pain Point: Disruptive workflow interruption caused by AI coding agents. Developers must frequently alt-tab or window-switch to a terminal to grant permissions or respond, breaking focus and slowing development velocity.
2. Target Audience: The primary user personas are professional software engineers, full-stack developers, and technical founders who regularly use AI coding assistants like Claude Code for daily development tasks, often managing multiple projects or agent sessions concurrently.
3. Use Cases: Essential for developers who are pair-programming with an AI agent while simultaneously researching documentation or reviewing PRs in the browser; for those managing long-running agent tasks (e.g., refactoring) while browsing; and for users of minimalist terminals that lack built-in desktop notification support for agent prompts.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros: Dramatically reduces context-switching overhead. Provides a delightful, at-a-glance overview of all active AI sessions. Privacy-focused, local execution aligns with developer security expectations. One-time pricing model offers long-term value versus subscriptions. Non-blocking hooks ensure agent performance is not degraded.
   • Cons: Limited to Chromium-based browsers (Chrome, Arc, Brave, Edge). Requires initial manual extension load via Developer Mode, which may deter non-technical users. Linux/Windows users lack auto-start daemons, requiring manual command execution (npx getagentpets start). Only supports three specific agent platforms (Claude Code, OpenCode, omp).

2. Key Alternatives & Differentiation:

   • Native Terminal Notifications: iTerm2, Ghostty, or Kitty have built-in support for Claude Code notifications. Differentiation: Agent Pets provides a unified, visual interface across all agents and works identically regardless of terminal emulator, with interactive buttons directly in notifications and browser.
   • Custom Script Hooks: Developers can write their own shell scripts or use tools like terminal-notifier to create custom alerts. Differentiation: Agent Pets offers a polished, productized solution with persistent visual pets, a management dashboard, and cross-tab reply functionality without any scripting or maintenance required.
   • Cloud-Based Agent Dashboards: Some AI coding platforms offer web-based dashboards. Differentiation: Agent Pets is entirely local, ensuring zero latency, no data transmission, and functionality that works offline or with local models.

Frequently Asked Questions (FAQ)

1. Does Agent Pets work with GitHub Copilot or Cursor? No, Agent Pets currently has built-in integrations only for Claude Code, OpenCode, and the omp agent framework. It does not support GitHub Copilot, Cursor, or other AI coding assistants that use different underlying APIs or notification systems.
2. Is the Agent Pets Chrome extension safe and secure to install? Yes, from a privacy perspective. The extension is loaded unpacked from a local directory (~/.agentpets/extension) and only communicates with the helper daemon on your computer (127.0.0.1). It requires no broad host permissions, and its operation is transparent and local. Security-conscious users can audit the code installed via npm.
3. Can I use Agent Pets on multiple computers with one license? The license is intended for one user. The policy allows activation on up to 3 browsers, which typically means across a primary and secondary machine for a single developer. It is not a site-wide or team license for simultaneous use across multiple users' machines.
4. What happens if the local helper daemon crashes or is stopped? If the getagentpets helper process stops, the extension will lose its connection to the coding agents. The pets will disappear, and you will no longer receive notifications or be able to approve requests from the browser. You must restart the daemon (npx getagentpets start on Linux/Windows; it auto-restarts on macOS) to restore functionality.
5. How does the "Your Cat" personalization feature work technically? The "Your Cat" feature uses the HTML Canvas API client-side in the extension to crop and resize your uploaded image. The processed image data is stored locally in your browser's extension storage (likely chrome.storage.local). It is never uploaded to any server; all customization occurs locally on your device.