token-watch
A cache-temperature tube and plan limits that say when to send, clear or switch models.
Illustration. Actual results depend on the mod, terminal and environment.
Install
/plugin marketplace add arviaja/token-watch
/plugin install token-watch@token-watch
/reload-pluginsRun these commands one at a time in a Claude Code session.
claude plugin marketplace add arviaja/token-watch
claude plugin install token-watch@token-watchThen run /reload-plugins in your session, or start a new session.
Please install the Claude Code mod "token-watch" (https://github.com/arviaja/token-watch). Before installing, read the Security and Requirements sections of its README and tell me what it reads and writes, which commands it runs, whether it uses the network, and which Claude Code version, OS and terminal it needs. If everything looks fine, run claude plugin marketplace add arviaja/token-watch and claude plugin install token-watch@token-watch, then remind me to run /reload-plugins in my session.Send this to Claude Code so it reads the author’s instructions before installing.
git clone https://github.com/arviaja/token-watch
cd token-watch
claude --plugin-dir .Clone the repository and try it in a local session with --plugin-dir.
What it accesses
Based on the author’s documentation, checked when listed; use claude plugin validate to check for yourself.
Reads
YesReads per-request token counts and model, plan limits, /cost, repo folder name and context breakdown.
Writes files
YesWrites per-session usage snapshots and the band setting to the mod store (~/.claude/plugins/store), kept 8 days.
Runs commands
UnspecifiedNot stated explicitly in README; no process launches seen in code.
Network
YesOnly /token-watch recommend calls a model, after you confirm, with the pane's usage data.
Introduction
token-watch is a Claude Code mod that shows the token use, plan limits and cache temperature of the sessions on this computer, live, in the CLI and the desktop Code tab. The author states it only observes: it does not change, block or delay a request.
- Band: a cache tube that empties from red to blue over the cache life (1 hour or 5 minutes), the stage (LIVE, HOT, WARM, COOLING, COLD) with minutes left, and the cost to re-warm.
- One action at a time: e.g.
5h full at 15:31: pause or use Sonnet,send now: after 14:32 the next message costs $2.40,/clear if the topic changed. - Limits: weekly and 5-hour use with a projection at the pace so far; with an API key, today's cost and $/h instead.
- Pane (
/token-watch): Now, Session, Week, Why and Help tabs with cost by model and scope, cache-write causes, the week over time and the context breakdown. /token-watch recommend: after you confirm the cost, sends the pane's usage data to a model for token-saving advice.
The cache life is read from the /cost figures. Requires Claude Code 2.1.287 or later.
Who it is for
- Subscribers who want to know when the cache goes cold and whether limits will run out
Requirements
- Claude Code ≥ 2.1.287
- The cache life needs an exact price for the model in its table; otherwise it shows cache life unknown.
- recommend costs one model call, which counts against a subscription's allowance.