a 𝟱𝟵 𝗽𝗮𝗴𝗲 𝗱𝗮𝘀𝗵𝗯𝗼𝗮𝗿𝗱 𝗱𝗲𝘀𝗶𝗴𝗻 𝗴𝘂𝗶𝗱𝗲, 7 color palettes, Power BI templates, a requirements worksheet, a prototyping tool and a full Excel checklist. Japan's Digital Agency released the whole pack for free. Most files are in Japanese. Hand the whole pack to Claude Code or Codex and turn it into a reusable 𝗗𝗮𝘀𝗵𝗯𝗼𝗮𝗿𝗱 𝗗𝗲𝘀𝗶𝗴𝗻 𝗦𝗸𝗶𝗹𝗹. Once it's a Skill, Claude Code or Codex can build or review dashboards against specific rules for requirements, layout, charts, and color. You don't have to rely on vague prompts like more professional or less AI-looking. The guide splits the work into 𝘁𝗵𝗿𝗲𝗲 𝘀𝘁𝗮𝗴𝗲𝘀: define requirements, build a prototype, then implement and review. Each stage comes with its own tool. The requirements worksheet asks who will use the dashboard, why they need it, what they need to know, what they should do next, and which data and constraints already exist. Next, turn those answers into a prototype and show it to actual users and the project team. Use their feedback to revise it before building it in Power BI. Split a 𝟭𝟲:𝟵 𝗰𝗮𝗻𝘃𝗮𝘀 into 2 to 6 rows and columns. Put the overview and most important metrics in the top left. Put the details that support decisions and actions in the bottom right. 𝗣𝗶𝗰𝗸 𝗰𝗵𝗮𝗿𝘁𝘀 𝗯𝘆 𝘁𝗮𝘀𝗸: → trends over time: line chart → compare quantities: bar chart → composition: area chart or pie chart Bar charts start at zero. Keep each legend next to its chart. Use 1 to 5 colors per chart. The 𝟳 𝗽𝗮𝗹𝗲𝘁𝘁𝗲𝘀 are Blue, Light Blue, Cyan, Green, Orange, Red, and Solid Gray. Each one includes colors and contrast ratios for text, backgrounds, charts, and increase or decrease states. You don't have to tweak every chart separately to make the dashboard look consistent. The 𝗘𝘅𝗰𝗲𝗹 𝗰𝗵𝗲𝗰𝗸𝗶𝘀𝘁 can become an agent review pass: → can you see the most important numbers first? → does the sort order have a clear meaning? → are charts and legends next to each other? → have unnecessary gridlines, borders, and decorations been removed? → are chart titles and data types clear? → do bar charts start at zero? → is the data's last updated date visible? → do display, speed, and interactions work on real devices? The guide repeats three checks throughout: can users find what they need quickly, will any chart mislead them, and can they make a decision or take action afterward? The official page says the pack is for government agencies, public institutions, companies, and individuals. The GitHub repo also includes JSON themes and map data for Japan's prefectures and municipalities. The Power BI templates and JSON themes use 𝗣𝗗𝗟 𝟭.𝟬. Attribution isn't required when you modify them for your own dashboard. If you republish them unchanged, you need to credit the source. English page + all downloads: GitHub: