Claude Pet

What is Claude Pet

Claude Pet is a Claude Code companion built for developers who leave their terminal running and do not want to miss what happened next. It listens for Claude Code hook events, then turns those events into sounds, visual reactions, memory notes, and break reminders. Because it is a local Claude Code companion, it is aimed at people who want better feedback while coding without sending their workspace history to a cloud service.

The app is presented as free and MIT licensed, with local storage on your machine. It works with macOS, Windows, and Linux, and its core job is simple: keep Claude Code visible, keep your project context organized, and keep your body from staying still for too long.

Key features

Claude Pet combines several functions into one Claude Code companion:

• Hook-based alerts: it reacts to real Claude Code hook events, so you can hear and see when Claude is reading, editing, running bash commands, succeeding, failing, or asking for attention.
• Project memory: it stores decisions, conventions, fixes, and gotchas in a local SQLite graph, then injects a ranked recap of up to 800 tokens into a new session.
• Self-learning skills: repeated patterns can be promoted into Claude Code SKILL.md files under ~/.claude/skills/, with levels that progress from hatchling to master.
• Ergonomics coach: it counts real work time and reminds you to take eye breaks every 20 minutes, chin tucks every 30 minutes, and wrist circles every 45 minutes, with stretch demos you can mirror.
• GitHub repo watcher: it can monitor public or private repositories for commits, PRs, reviews, releases, CI runs, deploy status, and issues.
• Local-first storage: memory lives in ~/.claude/claude-pet/memory.sqlite, and the app says it uses 11 regex patterns to scrub sensitive data before writing.
• Dashboard view: clicking the pet opens a HUD-style dashboard with memory and ergonomics data, including graph and gauge-style visuals.

This Claude Code companion is also opinionated about quiet operation. It deduplicates events by GitHub event ID, uses ETag-cached polling, and avoids replaying the same recent GitHub activity when you first add a repo.

How to use it

1. Download the latest release from the project page or GitHub.
2. Install Claude Pet on your desktop system and let it run alongside Claude Code.
3. Let the app observe Claude Code hook events so it can react when work changes state.
4. Add a project so Claude Pet can build local memory about the decisions and fixes you make.
5. Optionally add GitHub repositories to watch commits, pull requests, reviews, releases, CI runs, deploys, and issues.
6. Keep the dashboard handy if you want to review the local memory graph or check break timing.

A practical way to use this Claude Code companion is to leave it active during longer coding sessions. The pet becomes most useful when your terminal is busy, your browser is closed, and you still want a reliable signal that something important happened.

Who it is for

Claude Pet is for developers who use Claude Code and want more than plain terminal output. It fits:

• solo builders who jump between many projects and need project memory
• teams or maintainers who want repository alerts without setting up webhooks
• people who like subtle desktop feedback instead of checking a browser tab
• developers who want a built-in ergonomics coach during long sessions
• anyone who prefers a local Claude Code companion over a cloud service

It also makes sense for users who work across multiple repositories and want one place to watch reviews, deploys, and CI results. If you already spend hours in Claude Code, the app is meant to reduce context re-explaining and reduce missed status changes.

What to know before you use it

Claude Pet is a useful Claude Code companion, but it is not a replacement for GitHub itself or for your own judgment. It polls GitHub rather than using webhooks, so it depends on GitHub API limits: the project states 60 requests per hour without authentication and 5,000 per hour with an optional personal access token stored locally in ~/.claude/claude-pet/config.json.

It also does not claim perfect understanding of every event. Alerts are tied to supported GitHub event types and Claude Code hooks, so edge cases outside those signals may not be covered. The local memory is only as good as what the app captures from your sessions, and the ranked recap is limited to 800 tokens.

For sensitive work, the local-first design is reassuring, but you should still review what gets written to disk and rely on official GitHub pages for final source-of-truth checks on releases, deploys, and repository history. That is especially true if you need audit-level accuracy. As a Claude Code companion, Claude Pet is best used to stay informed faster, not to replace the underlying systems.

If you want a free, local Claude Code companion that combines alerts, memory, ergonomics, and GitHub monitoring, Claude Pet is built for that workflow and keeps the experience close to the terminal while staying out of the way.