wtf is claude-wake?

claude-wake solves a specific annoyance for people using Claude Code inside VS Code: when a usage limit stops a session overnight, nobody is around to type continue once the limit resets, so hours of working time are lost. claude-wake watches for that reset, brings the exact stopped panel back to the front, and automatically sends a resume message, so the work picks back up while the user is away or asleep. It runs two ways that share the same underlying engine and state: a zero dependency command line tool built for Node 18 or newer, and an optional menu bar app built with Tauri, which combines Rust with plain HTML rather than a full web frontend. Both can be installed from npm or built from source, and only one of the two should run at a time since they share the same state folder. Under the hood, claude-wake works in three steps. It detects a usage limit by tailing Claude Code's own session transcript files, which record the reset time along with a timezone, without doing any screen scraping. At reset time it uses a VS Code deep link to bring the exact panel back into focus, even if many panels are open. Since the panel's input ignores fake keystrokes but accepts a paste, claude-wake puts the resume message on the clipboard and performs a paste plus Enter, only after confirming the editor window is in front. It fires once per limit, skips a session that was already resumed manually, and its default message is written to be harmless even if it triggers when it was not actually needed. On macOS, sending a paste requires Accessibility permission, which claude-wake requests only for a small dedicated helper app it compiles locally, rather than for the terminal or Node itself. The menu bar app adds a popover with a live countdown to the next resume, the ability to edit the resume message, and pause and resume controls, while the CLI exposes commands like watch, install, status, pause, resume, doctor, test, and logs.