Skip to content
pacewick / local

Pacewick shows your Claude Code usage from the session logs already on your computer. Drop the folder and you get the current 5 hour block, your burn rate, the week, and the cost per day, model and project. Everything runs in your browser, so nothing is uploaded.

By The Pacewick team (Product and testing). Last updated .

~/.claude/projects (demo)
Current 5 hour block
35.0M1 h 31 min left
64% of the largest block, 207K/min
Today
$13
7 days
$159
Cache reads
97.7%
Opus 5.5$181
Sonnet 5.5$62
Haiku 5.5$0.77
Made up demo projects, real parser output.
01 / dashboard
Drop your .claude/projects folder here
Read on this device by your browser. Nothing is uploaded.
02 / steps

How to read your Claude Code usage in a minute

  1. Open the projects folder inside .claude in your home folder. On Windows it is %USERPROFILE%\.claude\projects.

  2. Drag the whole folder onto the dashboard, or press Choose folder.

  3. Then wait a few seconds while your browser reads the logs on this device.

  4. Finally, read the current block first, then today, the week and the cost per day.

Claude Code writes every session to a JSONL file, and each answer carries a usage record. However, one answer is written as several lines with the same usage. So the dashboard counts each answer once, by its message and request id, before it adds anything up.

As a result, the totals match what Claude actually did. For example, on one developer's real logs the parser read 955,052 lines and skipped 156,647 repeated ones, which would otherwise have doubled the count.

03 / the log

What the dashboard reads from one log line

Only the usage part of each answer matters here. It holds four kinds of tokens: input, output, cache writes and cache reads. In addition, the line carries a time, a session id and the folder you worked in.

Prompts, code and tool output sit on the same lines, but the dashboard never parses them. In fact, it skips every line that has no usage record at all.

invoice-api/5f2c.jsonl
{"type":"assistant","timestamp":"2026-11-07T10:38:12Z","sessionId":"5f2c...","cwd":"/Users/demo/code/invoice-api","message":{"model":"claude-opus-5-5","usage":{"input_tokens":12,"cache_creation_input_tokens":4810,"cache_read_input_tokens":402315,"output_tokens":1184}}}
read: usage, model, timestamp, cwdskipped: prompts, code, tool output
04 / panels

Your Claude Code usage, one panel at a time

These are real panels from our demo log: 12 work days, 3 made up projects and 651 million tokens.

  1. Claude Code usage of the current 5 hour block: 35.0M tokens, 64 percent of the largest block, 1 h 31 min left
    Block

    01The open block

    35.0M tokens with 1 h 31 min to go

    First, the panel you check most. The open 5 hour block holds 127 requests so far, which is 64 percent of the largest block in the log. The burn rate says where it lands by the reset.

    Learn more about 35.0M tokens with 1 h 31 min to go
  2. Claude Code usage per day as API price bars, peak day $36.18
    Days

    02Per day

    Busy days stand out, the peak is $36

    Next, the bars show the API price per day. So a heavy refactor day is easy to spot, and you can match it to the work you did.

    Learn more about Busy days stand out, the peak is $36
  3. Claude Code usage per model: Opus 5.5 $181.47, Sonnet 5.5 $62.39, Haiku 5.5 $0.77
    Models

    03Per model

    Opus 5.5 is 74 percent of the bill

    The model table splits tokens, output and price. For example, Haiku 5.5 did 181 requests for under a dollar, while Opus 5.5 carried most of the cost.

    Learn more about Opus 5.5 is 74 percent of the bill
  4. Claude Code usage per project: invoice-api, marketing-site and data-scripts with tokens and API price
    Projects

    04Per project

    Every repo gets its own line

    Because each log sits in a folder named after the project, the dashboard can split the totals by repo. That answers the question a client invoice or a team lead will ask.

    Learn more about Every repo gets its own line
  5. Claude Code token usage split: 97.7 percent cache reads, then cache writes, output and input
    Cache

    05Token types

    97.7 percent are cache reads

    In fact, almost every token is a cheap cache read of context Claude already has. That is why a huge token count can still mean a modest API price.

    Learn more about 97.7 percent are cache reads
  6. Claude Code usage CSV report with rows per day, per model and per project
    CSV

    06CSV report

    Totals to a spreadsheet, nothing else

    Finally, the report turns the same totals into a CSV file. Project names are hidden unless you tick the box, and prompts never leave the page.

    Learn more about Totals to a spreadsheet, nothing else
your turn

Read your own logs now

First, drop the projects folder on the dashboard. Then compare your week with the demo.

Claude Code usage dashboard on the demo log: current block, today, week, per day bars and per model table
Demo log: 12 days, 3 made up projects, 7,466 lines, read in 35 ms by the parser that runs in your browser.
05 / example

Claude Code usage on a real log

Here is the dashboard on our demo log, read by the same code that runs in your browser. Notice how little of the bill is output. Because 97.7 percent of the tokens are cache reads, the API price stays far below what the token count suggests.

2,483
requests counted
2,500
repeated lines skipped
$244.63
at API list prices
06 / prices

API list prices behind your Claude Code usage cost

API list prices per million tokens (claude.com/pricing, checked 7 November 2026)
ModelInputOutputCache write 5mCache write 1hCache read
Fable 5.1$10$50$12.50$20$0.25
Opus 5.5$4$20$5$8$0.20
Opus 5 and 4.5 to 4.8$5$25$6.25$10$0.50
Sonnet 5.5 and 5$2$10$2.50$4$0.20
Haiku 5.5$0.10$0.50$0.125$0.20$0.01
07 / guides

Guides to Claude Code usage, limits and cost

Each guide answers one question and opens the same dashboard, so start with the one you have today.

08 / pricing

Pricing

The dashboard is free, with no account and no limit. Only the CSV report counts: three a day are free, and credits cover more. See the pricing page for details.

Free
No account
$0/ every day
  • The full dashboard, always
  • 3 CSV reports a day
  • Logs stay on your device
  • Every model and project
Start free
Report pack
25 reports, never expire
$3.99/ one time
  • Full CSV every time
  • No daily limit
  • Credits never expire
Best value
Starter yearly
25 reports every month
$29/ per year
  • About $0.10 per report
  • Full CSV every time
  • Priority support
Starter monthly
25 reports every month
$4.99/ per month
  • Cancel anytime
  • Full CSV every time
  • Priority support
Team
500 reports every month
$12.99/ per month
  • Reports for a whole team
  • Priority queue
  • Everything in Starter

1 credit = 1 CSV report. The dashboard itself is always free. Pack credits never expire. Payments go through Lemon Squeezy, the merchant of record.

honest limits

What this Claude Code usage dashboard cannot see

Your plan limit also counts chats on claude.ai, the desktop app and the phone app. Those chats are not in Claude Code logs, so the dashboard cannot see them. If you chat a lot, the real limit may be closer than the gauge shows.

Also, Anthropic does not publish the limits in tokens. That is why the gauge compares the open block with your own largest block. For the official percentage, run /usage in Claude Code or open Settings and then Usage on claude.ai.

Finally, Claude Code deletes session files older than 30 days by default. So if you want a longer history, raise cleanupPeriodDays in your settings first.

09 / questions

Claude Code usage questions

How do I check my Claude Code usage?

Run /usage in Claude Code for the official plan bars. For history, cost per day and a per project split, drop your .claude/projects folder on the dashboard on this page.

Is this Claude Code usage dashboard free?

Yes. The dashboard is free with no account. Three CSV reports a day are free as well, and credits unlock more.

Do you upload my logs?

No. Your browser reads the files on your own computer. Only the optional CSV report sends totals to our server, and never prompts, code or file paths.

When does the Claude Code 5 hour limit reset?

The window starts with your first message and resets in full 5 hours later. The dashboard shows that countdown for the block that is open now.

Is Pacewick made by Anthropic?

No. Pacewick is an independent tool. It reads the log files that Claude Code writes on your machine and uses public API list prices.

Know your pace before the limit does

The dashboard is free and your logs stay on this device. Three CSV reports a day are free too.

Block64% of peak
Projection at reset53.9M

Sources and references