ReRead
See how much of your coding agent's usage was re-reading context.
ReRead is a token usage and prompt cache analyzer for Claude Code, Codex and Antigravity. It reads the session files these agents already keep on your computer and shows how much of your usage went to re-reading the conversation from the cache, writing it again after the cache was lost, and thinking. In VS Code it also counts down until your Claude Code prompt cache goes cold.
Live on the VS Code Marketplace, Open VSX and npmInstall from the Marketplace Open VSX, for Cursor and others npm: reread-cli
What it does
Out of your way.
- A status bar countdown until the prompt cache of your latest Claude Code session goes cold, with what the next reply will cost if it does
- A report of how much of your usage was re-reading history, writing it again, thinking, and everything else
- When and why the cache was lost, such as a long break, a model switch or a change near the start of the context
- Your costliest sessions, and cost by project, model and effort level, at API prices
- Also a command-line tool, npx reread-cli, with a status line for Claude Code
- Reads only token counts, models, times, folders and branches; never what you or the agent wrote
Quick facts
Before you install.
- Price
- Free
- Works offline
- Yes, no connection needed
- Account
- None needed
- Platforms
- VS Code, Cursor, Antigravity and other VS Code based editors; any terminal with Node.js 20 or newer
- Status
- Out now
Questions
Asked and answered.
- Does ReRead work offline?
- Yes, ReRead works without an internet connection.
- Do I need an account to use ReRead?
- No. There is no sign-up; just install it and start.
- Where can I get ReRead?
- ReRead is free on the VS Code Marketplace (https://marketplace.visualstudio.com/items?itemName=hamperpick.reread); on Open VSX, for Cursor, Windsurf and VSCodium (https://open-vsx.org/extension/hamperpick/reread); and on npm as reread-cli (https://www.npmjs.com/package/reread-cli).
- Why does Claude Code cost more after a break?
- Every reply sends the whole conversation to the model again. While the prompt cache is warm, that history is read from the cache at a small fraction of the input price. The cache lasts 5 minutes after the last reply, or 1 hour when a session uses the longer cache. After that, the next reply writes the whole context to the cache again, at 1.25 times the input price (2 times for the 1-hour cache). ReRead's status bar shows how long the cache stays warm, so you can compact or start a new session before it goes cold.
- Does ReRead upload my conversations?
- No. ReRead makes no network requests. It reads the token counts, model, time, ids, folder and git branch of each reply from the session files on your computer, and never keeps or shows what you or the agent wrote.
- Which coding agents does ReRead support?
- Claude Code, Codex and Antigravity. The status bar countdown is for Claude Code; the report covers all three. Models ReRead has no price for, such as DeepSeek or Qwen models, are counted but not priced.
- Something else?
- Write to contact@hamperpick.in.