Where did 🌊
my tokens go?

Hit your usage limit by lunch? Wake reads the transcripts Claude Code keeps on your Mac and shows you what filled your context, and what would have sent less.

Wake listing Claude Code sessions by project, in front of the Usage window with cost per day

Find out why it got expensive

Every reply re-sends the whole conversation, so a long session keeps paying for its first turns. The Usage window groups tool calls by what the agent was doing (reading files, searching, running commands) and gives you a tip on what would send less. Export it as a Markdown checklist and hand the fix back to your agent.

The Usage window grouping a week of tool calls by intent, with the most expensive shell commands listed below

Replay any session

Read a session as a timeline: prompts, answers, commands with their output, file edits as diffs and subagents nested where they ran. Each step shows its time, tokens and how much it added to the context. Press Cmd+F to search every session at once and jump right to the match.

A session timeline with a shell command expanded to show its output and what it added to the context
Wake listing Claude Code sessions by project, in front of the Usage window with cost per day

Look back along the trail

Wake only reads your transcripts. It never changes them, never goes online and never collects anything.

Download for macOS

Free and open source, built with Xojo. Signed and notarized.