Guide · Agent skill
How to use the Claude Code Usage Receipts agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The receipts skill is an agent skill for Claude Code that writes a personal usage and impact report for one developer, built from the session transcripts already saved on your machine. It is useful if you need to explain what Claude Code has been used for, for example to a manager or in a self-review. This guide covers what the skill does, how to install it, how to trigger it with plain requests, and what it needs and touches on your computer.
Claude Code Usage Receiptsanthropics/claude-plugins-official · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Download the receipts skill ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project.
- 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 4Check Node.js is installed. The skill runs a bundled Node.js script, so Node.js must be available on your machine.
- 5Ask for a report. Say something like "make me a receipts report for the last month" or run /receipts.
- 6Open the outputs. Find the markdown report and the HTML receipt in your home directory, and review them before sharing.
What the receipts skill makes the agent do
When you ask for a receipts or usage report, the agent runs a local script that reads your Claude Code session transcripts in ~/.claude/projects. It summarizes sessions, prompts, active days, files touched, and roughly how much of your usage went to each project. It also cross-checks activity against your local git history.
The output is a markdown report plus a self-contained, styled HTML receipt written to your home directory. The HTML page includes an Export CSV button for the by-project table. Most of the processing is the local script; the only model call is one final write-up.
Install the receipts skill
Download the ZIP from the skill's page on this directory and unzip it. There are two places to put the skill folder:
Option A is your personal skills folder, which makes the skill available across projects. Option B is a project's .claude/skills/ folder, which makes it available only in that project. You can also upload the skill in the Claude apps under Settings → Capabilities → Skills.
- Personal: unzip into ~/.claude/skills/
- Single project: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the skill
Trigger the skill with natural requests
You do not need a special syntax. Ask in plain language, and the agent will match your request to the skill. The skill description lists phrases such as "receipts", "impact report", "usage report", "show my Claude Code activity" and "prove the value of Claude Code". You can also run /receipts directly.
You can name a period in your request. The skill understands week, month (the default if you give none), quarter, year, or a bare number of days. You can also name a project, and the report will be scoped to matching projects, totals included.
- "Make me a receipts report for the last month"
- "Give me a usage report for the past quarter"
- "Show my Claude Code activity for the last 14 days"
- "/receipts year"
- "/receipts for anthropic"
What you get back
The agent reports the period it covered and writes two files to your home directory: a markdown report and an HTML receipt. The report is organized around your own work with Claude Code, mapped to the project where that work landed. Claude Code's own internal files and the ~/.claude folder are excluded.
Spend is shown only as a share of total relative compute per project. The skill does not break spend down by activity type, because that would be a modelling choice rather than a measurement.
- Totals for sessions, prompts, active days, files touched and lines touched
- Per-project breakdown with a share of spend
- Commit counts matched against git, where a git repo is available
- A research bucket for sessions that used the web, Slack or a dashboard without touching files
What the skill needs and what it touches
The skill needs Node.js, because it runs the bundled mine-transcripts.mjs script. Git is optional and is used only to cross-reference local commits. The skill needs local Claude Code session transcripts in ~/.claude/projects to have anything to report on.
Safety notes: the skill runs scripts and reads every transcript file in the chosen window, so it will see your past prompts and project paths. It does not use the network. Commit matching is approximate, and spend is estimated per project rather than measured. Because it only sees one machine, it cannot produce an organisation-wide report.
- Requires Node.js to run the bundled script
- Git is optional, used for local commit cross-referencing
- Runs scripts: yes
- Uses network: no
- Reads local transcripts, including past prompts and project paths
- Covers one machine only
Tips for better reports
Pick the period that matches the question. A week is quick and useful for check-ins, while a quarter or year suits self-reviews. Narrow the scope with a project name if your manager asks about one codebase.
Treat the numbers as estimates for justification, not as an invoice. Read the report before sharing it, since it includes project paths and prompt-derived details.
- Use a project name to scope the report to one codebase
- Expect commit matching to be approximate
- Check the HTML receipt in a browser before sharing it
What people use it for
- Justifying Claude Code usage or spend to a manager with a per-project breakdown
- Preparing a self-review or performance check-in with a summary of your activity
- Checking what you have been using Claude Code for over the past week, month or year
- Showing the by-project table as a CSV using the Export CSV button in the HTML receipt
Frequently asked questions
Is the receipts skill free?
This page does not state a price or licence terms for the skill itself beyond its listed licence, Apache-2.0. The skill runs locally and makes no network calls, but the Claude Code usage it reports on is your own.
What do I need to run the receipts skill?
You need Node.js to run the bundled mine-transcripts.mjs script, and local Claude Code session transcripts in ~/.claude/projects. Git is optional and only used to cross-reference commits.
How do I install the receipts skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.
Does the receipts skill send my transcripts anywhere?
The skill is listed as not using the network. It reads transcript files locally, and the only model call is one final write-up over a small JSON summary.
Can it show my team's or organisation's usage?
No. It only covers one machine's sessions, so it is a per-developer report and not an aggregated organisation-wide report.
Are the cost numbers exact?
No. Spend is estimated per project as a share of relative compute, and commit matching is approximate. Use the report for justification and self-review rather than exact billing.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use the receipts skill if you are a developer who needs a clear, local summary of your Claude Code activity for a manager, a self-review or a personal check-in. Pick something else if you do not want local transcripts scanned, need an organisation-wide report, or need exact cost or commit attribution.
Download the skill