Skip to content

Get from a CSV to a chart your boss can read, in one prompt

← Use cases Data

Get from a CSV to a chart your boss can read, in one prompt

Every team has a CSV nobody looks at — exports from some system, columns of numbers, no picture. Claude reads it and builds an interactive chart page you double-click open. No Excel wrangling, no BI-team ticket, no licence. The gateway use case — once this clicks, you'll see CSVs differently. Audience: beginner · ~15 min · 5 steps

You'll end up with

A self-contained `chart.html` — an interactive chart of your CSV with toggles for the dimensions that matter, plus a totals strip that proves the numbers match the file.

Before you start, you need
  • Any CSV — or nothing, Claude will invent a realistic one to practise on
  • Claude Code installed and a terminal open (see Day 1)

Somewhere in your team’s shared drive is a CSV that answers a question someone keeps asking. It stays unanswered because turning the CSV into a picture means Excel pivot tables, or a BI ticket with a three-week queue, or that one colleague who “does dashboards”.

This is the shortest use case in the hub, on purpose: one synthetic file, one , one chart. It’s the gateway — when this clicks, you stop seeing CSVs as homework and start seeing them as one prompt away from a picture.

Compliance check before you start. This walkthrough uses synthetic data. With real exports, the usual posture applies: aggregate counts and totals are normally fine on your workstation; row-level customer data (names, IBANs, balances) needs your line manager’s OK before it lands in any Claude-readable folder. When in doubt, ask Claude to aggregate first and chart the aggregate.

Step 01

Build the workspace

Open the .

Press ⌘+Space, type “Terminal”, and press Enter.

Open the Start menu (press the Windows key), type “Ubuntu”, and press Enter. If you don’t see Ubuntu listed, install WSL first.

In Ubuntu, ~/Desktop is a folder inside ’s Linux home (/home/<your-Linux-username>/Desktop) — not the Windows desktop you see in File Explorer at C:\Users\...\Desktop. That’s fine: the files are real and Claude can read and write them. Anywhere this use case says “open in Finder / File Explorer”, run explorer.exe . from your Ubuntu terminal — Windows opens that exact WSL folder in Explorer.

Type each line:

mkdir -p ~/Desktop/claude-lab/csv-chart
cd ~/Desktop/claude-lab/csv-chart
claude --dangerously-skip-permissions
  • mkdir -p ~/Desktop/claude-lab/csv-chart — make a folder inside claude-lab on your Desktop (-p creates claude-lab too if it’s not there yet — it’s the one home for all hub use cases).
  • cd ~/Desktop/claude-lab/csv-chart — move into it.
  • claude --dangerously-skip-permissions — start here. The flag stops Claude you for permission on every file write — safe in a fresh, dedicated folder like this one. (If you’d rather see every prompt for your first run, just type claude — same thing, more interruptions.)
Step 02

Ask Claude to invent a CSV worth charting

Have a real CSV? Drop it in the folder and skip to Step 3. Otherwise, tell Claude:

Create a file called data.csv in this folder: 24 months (2024-07 through 2026-06) of synthetic monthly transaction counts for a Greek retail bank, broken down by region and channel. Columns: month,region,channel,transactions. Regions: Attica, Macedonia, Thessaly, Crete. Channels: , app, web, atm. One row per month × region × channel combination.

Make the numbers tell a story: app transactions growing steadily and overtaking branch sometime in 2025; branch declining slowly everywhere but holding up better in Crete; a visible December spike in every channel; atm roughly flat. Plausible magnitudes for a retail bank — hundreds of thousands per month per region, not millions.

Claude writes the file. Notice you specified the story, not the numbers — that’s what makes synthetic data useful for practising: you know what the chart should show before you build it.

Step 03

One prompt, one chart

Paste this:

Read data.csv. Build a single self-contained file chart.html that I can open by double-clicking — no server, no internet, no external libraries; draw on a <canvas> with plain JavaScript.

  • A line chart of monthly transactions over time, one line per channel.
  • Toggle buttons for the regions — clicking “Crete” filters the chart to Crete; clicking “All” brings everything back.
  • Hovering a point shows the exact value in a small tooltip.
  • A totals strip above the chart: total transactions, busiest month, fastest-growing channel — recomputed live for whichever region filter is active.
  • Axis labels a non-data person can read, thousands separators on the numbers. Accent colour #007a8a.

Thirty to ninety seconds later, double-click chart.html in Finder / File Explorer. Click the region toggles. Hover the December spike. Find the month where the app line crosses the line — the story you ordered in Step 2, now visible.

Step 04

Make Claude prove the chart matches the file

A chart that’s 3% wrong is worse than no chart — it looks authoritative. Paste:

Verify your own chart. Recompute from data.csv with commands, independently of the page: the grand total of transactions, the total per channel, and the busiest month — show me the commands and their output. Then confirm the numbers embedded in chart.html match those figures exactly. If anything differs, fix the page and re-run the check.

Claude sums the actual file and compares it to what the page displays — arithmetic, not assurances. Same make-the--prove-it habit as every use case in this hub; it’s two minutes, and it’s the difference between “nice picture” and a number you’d repeat in a meeting.

Step 05

Now do it with your real CSV — and make the loop permanent

The synthetic run was calibration. The real run is: drop your actual export into the folder, then —

Read my file q2-export.csv and describe it: columns, row count, date range, anything odd. Then propose the two most useful charts for it and ask me which one to build.

Letting Claude propose the chart matters with real data — it has read the columns, and it often spots the dimension worth splitting by that you’d have flattened.

Then lock the loop in with CLAUDE.md — the magic filename reads automatically every time it starts in this folder:

Create a CLAUDE.md in this folder. Put in it my charting rules: output is always a single self-contained HTML file, canvas + plain JavaScript, no external libraries; always include interactive filters for any column with 2–8 distinct values; always include a totals strip recomputed for the active filter; always verify the page’s numbers against the CSV with commands before telling me it’s done; accent colour #007a8a; axis labels readable by a non-data person, thousands separators on numbers.

Next CSV — any CSV, any month: drop it in, claude --dangerously-skip-permissions, and say “chart this”. That’s the whole workflow. The BI queue is for the dashboards that need to live forever; this folder is for the question someone asked at standup.

Next use case →

Chain it together: complaints in, Monday briefing pack out

The capstone. You've done the heatmap, you've done the reply drafts — now chain them. One describes the whole Monday workflow, and from then on a single line ("do the Monday pack") turns a fresh complaints CSV into a themed report, reply drafts for the worst theme, an interactive dashboard, and a ready-to-send summary email. This is where stops being a tool and becomes a workflow.

Open it → Back to all 15 use cases