about mytokenuse.dev

A terminal-style dashboard for turning local LLM usage into short-lived, inspectable cost and token telemetry.

what it does

LLM usage, translated into numbers you can act on.

mytokenuse.dev collects sanitized ccusage reports from your machine and renders them as dashboards for model cost, total spend, token volume, cache behavior, and session-level outliers.

The value is practical: see which models drive spend, compare cost over time, find expensive sessions, and share a temporary snapshot when a teammate needs evidence instead of a screenshot.

input: local ccusage reports
output: private dashboard snapshot
default window: last 30 days
default visibility: private
default retention: 7 days
share mode: anyone with link

user guide

CLI first, dashboard second.

Install nothing permanent if you do not want to. Use npx mytokenuse, sign in once, publish a snapshot, then inspect it in the browser.

CLI guide

$npx mytokenuse publish

login: Sign in with GitHub and store a scoped publish token locally.
publish: Collect the last 30 days of ccusage reports, sanitize them, upload a private snapshot, and open the dashboard.
publish --dry-run: Preview collection and sanitization without uploading data.
publish --share: Publish the new snapshot as shared for anyone with the URL.
snapshots: List your stored snapshots with visibility, row counts, size, and expiration.
share / unshare: Switch a snapshot between shared and private visibility.
delete: Delete one snapshot and its stored usage data.
delete --all: Delete every stored snapshot for the active account.
logout: Revoke the active publish token and remove the local CLI config.

UI guide

Dashboard panels

Header: Wordmark, about link, account login/menu, and owner-only snapshot visibility control for switching private/shared access.
Totals rail: Model count, total cost, input tokens, and output tokens for the active snapshot.
Models rail: Ranked model cost list with proportional bars so the most expensive models stand out quickly.
Main panel: Tabbed chart and ledger area for model trends, cumulative spend, model totals, session trends, and session rows.

Main panel tabs

model trend: Stacked cost-by-model bar chart across the selected time grouping. Use it to see which models drive each day, week, or month.
running total: Cumulative cost-by-model line chart. Use it to see when spend accelerated and how each model contributes to the running total.
model ledger: Model table with total cost, input tokens, and output tokens. Use it for exact model-level totals instead of chart shape.
session trend: Bar chart of recent session costs, capped to the latest 80 sessions. Tooltips include session id, tokens, and models used.
session ledger: Session table with session id, agent, last activity, models, cost, total tokens, input, output, and cache token counts.

Controls

day / week / month: Available on model charts. Changes how periods are grouped without changing the underlying snapshot.
10s / 30s / 1m / 5m: Desktop cycle controls for dashboard-style use cases where the main panel rotates during passive monitoring.
private / shared: Owner-only snapshot control. Private requires the owner session; shared allows anyone with the URL to view the dashboard.

open demo

data privacy

Your data is used only to run this product.

Uploaded usage data is not sold, shared, used for training, or used for any purpose outside authentication, upload, storage, dashboard rendering, sharing controls, deletion, retention cleanup, and abuse prevention.

Once snapshot data is deleted, it is gone forever. The dashboard cannot restore deleted usage blobs.

Collected

Account: GitHub sign-in identity such as user id, name, email, avatar, GitHub login, session records, and publish-token metadata.
Usage snapshots: Sanitized ccusage daily and session reports: dates, model names, costs, token counts, cache token counts, agent label, last activity, model lists, and hashed session ids.
Snapshot metadata: Snapshot id, owner id, visibility, collector name and version, generated time, collection window, row counts, compressed sizes, creation/update timestamps, expiration, and deletion marker.
Upload metadata: Short-lived upload-session ids, scoped Blob paths, publish-token association, size limits, and upload expiration/finalization timestamps.

Retention

Usage snapshots expire after 7 days by default. Expired snapshots are cleaned up by the server, which deletes the stored daily and session report blobs and marks the snapshot deleted.

Unfinalized upload sessions are short-lived and are cleaned up when their upload window expires. Account and token records persist so login, ownership, rate limits, and revocation can work.

Delete

Single snapshot: Run mytokenuse delete <snapshot-id> or delete through the authenticated snapshot API.
All snapshots: Run mytokenuse delete --all. Use --yes only when scripting the irreversible confirmation.

Redaction

Local filesystem paths are redacted before upload. Session ids are hashed before they leave your machine, and the server rejects snapshots that still contain obvious local path strings.