Skip to content

borghei/Claude-Skills / product-team/design-direction

design-direction

Verified

Turns a product brief into a design-system spec: style direction, contrast-checked palette, type pairing, tokens, UX rules and chart choices. Use when starting a new product's UI, picking a palette or fonts, or auditing a token file.

references/charts.json

View on GitHub
{
  "version": 1,
  "note": "Data question -> chart. 'max' is the point where the chart stops working and the fallback takes over.",
  "charts": [
    {"id": "trend", "question": "How did a value change over time?", "keywords": ["trend", "over time", "time series", "growth", "history", "daily", "monthly"],
     "best": "Line chart", "alternatives": ["area chart for one series", "sparkline inside a table"],
     "avoid_when": "fewer than four time points (show the numbers) or no time axis", "max": "five series before it tangles; beyond that, small multiples",
     "a11y": "distinguish series by line style and direct labels, not colour alone; offer the data table"},
    {"id": "compare", "question": "How do categories compare in size?", "keywords": ["compare", "comparison", "categories", "versus", "by segment", "by region"],
     "best": "Bar chart sorted by value", "alternatives": ["grouped bar for two measures", "dot plot when bars are all similar"],
     "avoid_when": "the categories are ordered in time (use a line)", "max": "about 15 bars; more needs a horizontal bar or a searchable table",
     "a11y": "start the value axis at zero, label bars directly"},
    {"id": "rank", "question": "Who is top or bottom?", "keywords": ["rank", "ranking", "top", "leaderboard", "best", "worst", "top 10"],
     "best": "Horizontal bar chart, sorted", "alternatives": ["ordered table with inline bars", "slope chart for rank change between two dates"],
     "avoid_when": "differences are too small to see; show exact values instead", "max": "top 10-20 with an 'other' row",
     "a11y": "put the rank number in text, keep reading order equal to visual order"},
    {"id": "part-whole", "question": "What share does each part make up?", "keywords": ["share", "pie", "proportion", "percentage", "breakdown", "composition", "split", "part of whole"],
     "best": "Stacked bar (100%) or a single bar split into parts", "alternatives": ["donut with at most five slices", "waffle chart for one share"],
     "avoid_when": "parts do not add to a meaningful whole, or there are more than five parts", "max": "five parts; group the rest as 'other'",
     "a11y": "label each segment with name and percent; avoid adjacent hues that collapse under colour blindness"},
    {"id": "distribution", "question": "How are values spread?", "keywords": ["distribution", "spread", "histogram", "range", "outliers", "variance"],
     "best": "Histogram", "alternatives": ["box plot to compare groups", "strip or beeswarm for under 200 points"],
     "avoid_when": "the audience needs one headline number (give the median and range)", "max": "box plots for more than five groups",
     "a11y": "state the median and range in the caption"},
    {"id": "correlation", "question": "Do two measures move together?", "keywords": ["correlation", "relationship", "scatter", "versus", "impact of", "effect"],
     "best": "Scatter plot", "alternatives": ["bubble chart for a third measure", "binned heatmap for thousands of points"],
     "avoid_when": "one variable is time (use a line)", "max": "about 2,000 points before overplotting; then bin or use transparency",
     "a11y": "describe the relationship in words; do not imply causation in the title"},
    {"id": "funnel", "question": "Where do people drop off between steps?", "keywords": ["funnel", "conversion", "drop off", "pipeline", "steps", "checkout flow"],
     "best": "Horizontal bars per step with step-to-step conversion labelled", "alternatives": ["funnel shape only for presentations"],
     "avoid_when": "steps are not sequential or users can enter mid-way", "max": "about seven steps",
     "a11y": "write both the count and the percent of the previous step"},
    {"id": "target", "question": "Are we on track against a goal?", "keywords": ["target", "goal", "progress", "quota", "budget vs actual", "okr", "on track"],
     "best": "Bullet chart", "alternatives": ["progress bar with the target marked", "line with a target band"],
     "avoid_when": "there is no agreed target", "max": "one bullet per row; stack rows for many goals",
     "a11y": "state actual, target and percent in text; status needs a word, not just a colour"},
    {"id": "forecast", "question": "Where is this heading?", "keywords": ["forecast", "projection", "prediction", "future", "estimate"],
     "best": "Line with actuals solid, forecast dashed and an uncertainty band", "alternatives": ["fan chart for several scenarios"],
     "avoid_when": "the forecast has no stated method or interval", "max": "three scenarios",
     "a11y": "label where actuals end; describe the interval in text"},
    {"id": "anomaly", "question": "What looks unusual?", "keywords": ["anomaly", "outlier", "spike", "unusual", "alert", "deviation"],
     "best": "Line with an expected band and highlighted points outside it", "alternatives": ["control chart for process data"],
     "avoid_when": "there is no baseline to compare against", "max": "a handful of flagged points per view",
     "a11y": "mark anomalies with a shape as well as colour, and list them below the chart"},
    {"id": "hierarchy", "question": "How does a total break down through levels?", "keywords": ["hierarchy", "nested", "tree", "drill down", "categories within categories"],
     "best": "Indented table with bars, or a treemap for two levels", "alternatives": ["icicle chart", "sunburst only for exploration"],
     "avoid_when": "small parts matter (treemaps hide them)", "max": "two levels on screen at once",
     "a11y": "provide an expandable tree with the same numbers"},
    {"id": "flow", "question": "How does volume move between states?", "keywords": ["flow", "sankey", "movement", "migration", "transfers", "journey paths"],
     "best": "Sankey diagram", "alternatives": ["chord diagram for mutual flows", "table of from-to pairs"],
     "avoid_when": "there are more than about 30 links", "max": "two to three stages",
     "a11y": "offer a from-to table; do not rely on ribbon colour"},
    {"id": "geo", "question": "Where is it happening?", "keywords": ["map", "geographic", "region", "country", "location", "geo", "territory"],
     "best": "Choropleth for rates, proportional symbols for counts", "alternatives": ["sorted bar by region when location shape is irrelevant"],
     "avoid_when": "big empty regions dominate the visual weight; use a bar chart", "max": "one measure per map",
     "a11y": "provide the ranked region table; use a sequential palette that works in greyscale"},
    {"id": "heat-calendar", "question": "When does activity happen?", "keywords": ["calendar", "heatmap", "activity", "day of week", "hour of day", "streak", "intensity"],
     "best": "Calendar heatmap or day-by-hour grid", "alternatives": ["bar chart by weekday"],
     "avoid_when": "exact values matter more than the pattern", "max": "one year per view",
     "a11y": "sequential single-hue scale with a legend and tooltips that read the value"},
    {"id": "change-delta", "question": "What changed between two points?", "keywords": ["change", "delta", "before after", "difference", "variance", "bridge", "waterfall"],
     "best": "Waterfall for contributions to a change; slope or dumbbell chart for before/after", "alternatives": ["table with signed deltas"],
     "avoid_when": "there are more than two time points (use a line)", "max": "about 12 steps in a waterfall",
     "a11y": "use + and - signs and words, not only green and red"},
    {"id": "ohlc", "question": "How did a price trade within each period?", "keywords": ["ohlc", "candlestick", "price", "stock", "trading", "crypto chart"],
     "best": "Candlestick chart with volume below", "alternatives": ["line of closing price for casual users"],
     "avoid_when": "the audience is not trading; a close-price line is clearer", "max": "a few hundred candles per view",
     "a11y": "up/down by fill and outline, not only red/green; let users choose the colour convention"},
    {"id": "network", "question": "How are things connected?", "keywords": ["network", "graph", "connections", "dependencies", "relationships", "nodes"],
     "best": "Node-link diagram with filtering", "alternatives": ["adjacency matrix for dense graphs", "dependency list"],
     "avoid_when": "more than about 100 nodes are shown at once", "max": "100 nodes before it becomes a hairball",
     "a11y": "provide a list of each node's connections"},
    {"id": "realtime", "question": "What is happening right now?", "keywords": ["real time", "live", "streaming data", "now", "current load", "telemetry"],
     "best": "Rolling line with a fixed window and the latest value in text", "alternatives": ["status tiles with sparkline"],
     "avoid_when": "updates are faster than people can read; aggregate per second or minute", "max": "a fixed window (e.g. last 15 minutes)",
     "a11y": "announce threshold breaches politely, not every tick; allow pausing updates"},
    {"id": "single-kpi", "question": "What is the one number that matters?", "keywords": ["kpi", "headline number", "stat", "metric card", "big number", "scorecard"],
     "best": "Stat tile: value, unit, comparison period and change", "alternatives": ["stat tile with sparkline"],
     "avoid_when": "the number means nothing without context", "max": "four to six tiles per row",
     "a11y": "write the change as text ('up 4% vs last month'), not just an arrow"},
    {"id": "schedule", "question": "What happens when, and how long?", "keywords": ["schedule", "timeline", "gantt", "roadmap", "agenda", "bookings"],
     "best": "Gantt or timeline bars on a time axis", "alternatives": ["agenda list on small screens"],
     "avoid_when": "dates are uncertain; show quarters, not days", "max": "about 30 rows before grouping",
     "a11y": "list view with start, end and owner as a fallback"},
    {"id": "cohort", "question": "Do newer groups behave better than older ones?", "keywords": ["cohort", "retention", "churn", "signup month", "survival"],
     "best": "Cohort triangle heatmap", "alternatives": ["retention curves, one line per cohort"],
     "avoid_when": "fewer than four cohorts exist", "max": "about 12 cohorts",
     "a11y": "values printed in cells; single-hue sequential scale"},
    {"id": "uncertainty", "question": "How sure are we?", "keywords": ["uncertainty", "confidence interval", "error bars", "range", "margin of error"],
     "best": "Point with interval bars", "alternatives": ["gradient interval", "quantile dot plot for lay audiences"],
     "avoid_when": "the audience will read the bar ends as hard limits without explanation", "max": "about 20 estimates",
     "a11y": "state the interval and what it means in one sentence"}
  ]
}
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall