Claude Code usage monitor and dashboards: track usage and watch many sessions in one place

A Claude Code usage monitor answers "how much have I used and how close am I to my limit". The built-in answer is /usage plus the rate_limits fields your status line can show; community tools like ccusage and Claude Code Usage Monitor add history, charts and predictions. The other kind of dashboard answers "which of my sessions is working, waiting on me, or done": agent view (claude agents) does that for one machine, and a hosted board does it across machines and agents.

Usage monitors: tokens, cost and limits

Built in: /usage and the status line

/usage (also /cost and /stats) shows the current session's cost, your plan's usage limits and activity stats. On Pro, Max, Team and Enterprise plans it also breaks down what counts against your limits. For your whole account, see claude.ai/settings/usage.

To keep limits in view while you work, read them in your status line. Claude Code sends rate_limits.five_hour.used_percentage and rate_limits.seven_day.used_percentage to the status line script, for Pro and Max subscribers, after the first reply of a session:

#!/bin/bash
input=$(cat)
model=$(jq -r '.model.display_name' <<<"$input")
five=$(jq -r '.rate_limits.five_hour.used_percentage // empty' <<<"$input")
week=$(jq -r '.rate_limits.seven_day.used_percentage // empty' <<<"$input")
cost=$(jq -r '.cost.total_cost_usd // 0' <<<"$input")
out="[$model] \$$(printf '%.2f' "$cost")"
[ -n "$five" ] && out="$out | 5h $(printf '%.0f' "$five")%"
[ -n "$week" ] && out="$out | 7d $(printf '%.0f' "$week")%"
echo "$out"

Save it as ~/.claude/statusline.sh, make it executable, and point statusLine at it. The status line guide has the settings and more scripts. Note that cost.total_cost_usd is an estimate at list price, computed on your machine; on a subscription it is not what you pay.

ccusage

ccusage is an open-source CLI that reads the usage data coding agents keep on your machine and prints reports. It covers Claude Code and also Codex, OpenCode, Amp and Gemini CLI.

npx ccusage@latest daily

weekly, monthly and session give other cuts, and blocks shows Claude Code's five-hour windows. Nothing leaves your machine, and nothing runs in the background.

Claude Code Usage Monitor

Claude Code Usage Monitor is a live terminal dashboard with burn rate, predictions and warnings before you hit a limit. It uses the status line's official rate_limits when they are fresh and falls back to labelled local estimates.

uv tool install claude-monitor
claude-monitor

claude-monitor --once --output json prints one snapshot for scripts.

For teams

Team and Enterprise admins get an analytics dashboard at claude.ai/analytics/claude-code with active users, accepted lines, contribution metrics through a GitHub integration, and CSV export. API customers get usage and spend in the Claude Console. For per-user token and cost data in your own Grafana, Datadog or Honeycomb, Claude Code exports OpenTelemetry:

export CLAUDE_CODE_ENABLE_TELEMETRY=1
export OTEL_METRICS_EXPORTER=otlp
export OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4317

Session dashboards: many agents at once

Usage tells you what you spent. When you run several sessions, the more pressing question is which one is stuck waiting for you.

Agent view

Agent view is Claude Code's own dashboard for background sessions, in research preview as of October 2026. Dispatch work from the shell, then open the list:

claude --bg --name "flaky-test" "investigate the flaky checkout test"
claude agents

Sessions are grouped into Needs input, Working and Completed, each row with a one-line summary written by a Haiku-class model. Space peeks at a session and lets you reply, Enter attaches to it. Background sessions run under a supervisor process, so they keep going when you close the terminal, and each moves into its own worktree before it edits files. Scripts can read the same state with claude agents --json --all.

The limits: it covers Claude Code on one machine, sessions stop if the machine shuts down, and sessions you started in other terminals only appear once you background them. Ten parallel sessions also use your quota roughly ten times as fast.

Desktop app and claude.ai

The Claude desktop app shows parallel sessions side by side, each optionally in its own worktree. On claude.ai/code, cloud sessions and Remote Control sessions appear in one list you can open from your phone, and a Claude Code project shows which of its threads need you.

Claude Squad

Claude Squad is an open-source terminal app that runs several agents (Claude Code by default, plus Codex, Gemini, Aider and others) each in its own tmux session and git worktree, and lets you switch between them from one list. Install it with brew install claude-squad and run cs. It is a good fit if you live in the terminal on one machine.

Compare the options

A board for every agent and machine

Briefkin is the last row of that table, so here is exactly what it does and doesn't do. A one-command installer adds a hook to Claude Code, Codex, Cursor, OpenCode, Gemini CLI, Copilot CLI, Windsurf and Amp on each computer. Every session becomes a card that says whether it is working, waiting on you or done, with a one-line status written by the agent CLI you already have (so Briefkin never sees or pays for that model call) and the agent's own todo list as a checklist. You get a Telegram ping only when a session is waiting on you, and you can answer a local Claude Code session's question from the board or from Telegram.

It is not a usage monitor: it doesn't read your token counts or plan limits. Any script can post a number to it as a metric card with a history chart, though, so a nightly job that reports your spend lands next to your sessions. The docs cover the installer and the API, and the notifications guide shows the hook-only alternative if you just want pings.