Desoul Review Tiếng Việt

← Desoul Review

Skills

chart-review-rules

A Claude Code skill with about 70 rules for making and reviewing charts, dashboards and weekly reports. It asks who reads the page and what the data is before building, states only what the data shows, and leaves causes and actions to you.

Helps with

Before

Before: the title names the topic (Weekly Signups by Channel). Five same-weight number cards, a rainbow stacked bar chart with a legend, four coloured lines, then key findings that guess at causes (list fatigue, cadence, deliverability).

What the user typed

This is our weekly signups data by channel (signups.csv). Help me analyze it and create a dashboard. Save it as report.html.

After

After: one screen. The headline says signups are above the 1,000 goal but paid carries them while email falls. Four big number cards against the goal and the $10 limit, two charts with one accent each on grey and notes at 31 Aug and 10 Aug, then three short decision cards with their source. One yellow chip asks about a referral bonus.

What the user typed

This is our weekly signups data by channel (signups.csv). Help me analyze it and create a dashboard. Save it as report.html.

What the skill asked, and the answers

Claude: I read signups.csv: 10 weekly rows, 3 Aug to 5 Oct 2026, with new signups in 4 channels (organic, paid, referral, email). Correct me if I've misread any of it.

Claude: I can't tell the reasons from the file, so I need four answers before I build the dashboard. (1) Who will read this, and what will they decide afterwards? (2) Are these new signups per week, and where do they come from? Is the 5 Oct week a full week? (3) Was there a paid campaign or budget change around 31 Aug, a change to the email program, or a tracking change? (4) What counts as good here, such as a target or a normal range, and which levers can the reader actually pull?

You: That's right. (1) Me, the owner of a small online shop. I read it every Monday in about five minutes and decide next week's ad budget and email plan. (2) Yes, new signups per week, by the channel they came from. It's sample data, invented for a demo. The 5 Oct week is complete. (3) We raised the paid ad budget from $1,500 to $3,000 a week from the week of 31 Aug. On 10 Aug we cut the newsletter from two sends a week to one, to save time. Nothing else changed, and no tracking changes. (4) More signups is better. I want at least 1,000 a week in total, and a paid signup is worth it under $10. I can change the ad budget any week, go back to two newsletter sends, or start a referral bonus.

Download .zip

cp -r chart-review-rules ~/.claude/skills/

Ideas from Cole Nussbaumer Knaflic, Storytelling with Data (Wiley, 2015), in our own words. Nothing quoted. The book covers the call to action, the main point in one sentence and titles that state the point; the summary block, actions and warning flags are our additions.