Desktop pet for Claude Code that reacts to its activity
Sidecrab is a pixel crab desktop pet that sits on your screen and reacts in real time to what Claude Code is doing. It shows thought bubbles when Claude is thinking, throws claws up when a tool needs permission, and wanders when you step away. It solves the problem of lacking a visual companion for Claude Code, similar to the pet in ChatGPT's desktop app.
Entities
Related
Physical desk creature that mirrors AI agent status and accepts approvals via button press
A small hardware device that sits under a monitor and displays the status of AI coding agents (busy, needs approval, idle, celebrating) using colored lights. Users approve or deny agent tool calls by pressing the device, making agent interaction more tangible and engaging.
Claude status display on LilyGo T-Display S3 Long
A physical desk display that shows real-time Claude activity, including model, tool usage, elapsed time, token counts, context percentage, and usage bars. It solves the problem of needing to alt-tab to check Claude's status, providing a glanceable hardware monitor for users.
Rowboat: Open-source local-first work app with customizable surfaces on Claude
Rowboat is an open-source, local-first desktop application that extends Claude into a full-fledged work app. It allows users to build custom work surfaces for tasks like note-taking, reply suggestions, and more, using LLMs to assist in real-time. It targets professionals who need a more structured, customizable AI workspace beyond a chat interface.
Persephone: persistent agent toolbox for Claude Code
Persephone is a free, open-source dev notepad that gives Claude Code a persistent toolbox of reusable scripts. It solves the problem of Claude rewriting the same ad-hoc scripts every session by making scripts permanent and discoverable via a manifest.
NeatContext: lightweight desktop app to give LLMs domain knowledge for oncall incident handling
NeatContext is a desktop application that lets LLMs access domain knowledge to handle oncall incidents more accurately. It solves the problem of SRE agents lacking domain-specific context, enabling better incident response without heavy infrastructure.

