Desktop pets that spectate your coding agents

What it does

OpenPets is an Electron-based desktop app that parks a small animated pet on your screen via the system tray. It ships with bundled abilities—ambient check-ins, break reminders, focus timers, and playful actions—so it feels alive even without developer tooling. If you want, you can hook it up to Claude Code, OpenCode, or any MCP-capable agent so the pet shows decorative reactions and short speech bubbles during your session.

The interesting bit

The project treats coding-agent integration as ambient decoration rather than functionality: hooks map agent events like editing or testing to pet reactions, but they are explicitly best-effort, throttled, and drawn from static local message pools so prompts, file paths, secrets, and code never leak into bubbles. It is essentially a privacy-conscious glue layer between your terminal and a tamagotchi.

Key highlights

• Routes different agents or projects to their own dedicated pet windows via a pet-pack system.
• Ships an MCP server and first-party plugins for Claude Code, OpenCode, and experimental Pi support.
• Uses a local IPC discovery file with a per-run token; Windows/WSL setups can opt into TCP transport.
• Monorepo architecture splitting the Electron app, IPC client, MCP server, and agent-specific hook packages.
• Bundled abilities work out of the box; agent integrations are entirely optional.

Caveats

• Current desktop builds are unsigned, so macOS and Windows will likely throw security warnings on first launch.
• The Pi extension is explicitly marked experimental and awaits full CLI install validation.
• Generic shell activity is intentionally quiet by default, so don’t expect constant chatter from normal terminal use.

Verdict

Worth a look if you want ambient company during long coding sessions or you enjoy the novelty of agent-driven desktop pets. Skip it if you need signed binaries or you find system-tray mascots distracting.