← AI Color Picker - Design ToolCONTENT HISTORY

Update to AI Color Picker - Design Tool

Snapshot Oct 28, 2026 · 12:10 UTC · version 2.0.0

Collection source: downloaded plugin package.

WHAT CHANGED

First saved snapshot

No earlier snapshot is available to establish a change.

Save AI Color Picker - Design Tool and check future changes in Following.

Up to 3 plugins free · no card required.

Compare saved observations

Download comparison JSON
Full technical diff · 0 changed fields
Full snapshot data
{
  "description": "Assign stable colors and non-color cues to supplied chart series using their semantics, an actual chart background, and the Color Picker app. Use for a series-to-color mapping or revising confusing chart colors. Do not claim to render a chart, simulate color vision, or certify accessibility from color choices alone.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 223
    }
  ],
  "name": "assign-chart-colors",
  "skill_md_contents": "---\nname: assign-chart-colors\ndescription: Assign stable colors and non-color cues to supplied chart series using their semantics, an actual chart background, and the Color Picker app. Use for a series-to-color mapping or revising confusing chart colors. Do not claim to render a chart, simulate color vision, or certify accessibility from color choices alone.\n---\n\n# Assign chart colors\n\nRead [the picker contract and calculator reference](../reconcile-color-specs/references/color-tools.md) before tool calls or calculations. Use `../reconcile-color-specs/scripts/color_tools.py` for normalization and relevant background checks, resolving the path from this skill directory.\n\n## Establish series roles\n\n1. Collect series names, chart type, actual background, any fixed colors, and whether values are categorical, ordered, or diverge around a meaningful midpoint. Preserve existing series identities across revisions.\n2. For categorical data, propose distinct hue and lightness choices. For ordered data, use an ordered lightness sequence; for diverging data, use two contrasting ends and an explicit midpoint. Explain the mapping and keep fixed colors anchored.\n3. Establish how readers identify series: direct labels, legend entries, markers, dashes, or patterns. Include at least one usable cue besides color for every series; choose cues suited to the chart type.\n4. Clarify any required contrast target and which graphical object or text pairing it applies to. Use the [WCAG non-text contrast requirement](https://www.w3.org/TR/WCAG22/#non-text-contrast) of 3:1 for required graphical distinctions only where applicable; apply text targets to labels. Do not treat a background ratio as proof that adjacent series are distinguishable.\n\n## Select one series at a time\n\n1. Keep a working table in the conversation with series, proposed hex, cue, and selection status. Normalize proposals and fixed colors. Check each relevant color/background pairing with `contrast`; explain failing fixed colors and possible outlines or labels without inventing a rendered solution.\n2. Open `color-picker` for the current series with its proposed color, a small set of alternatives, and a `reason` naming the series. Keep labels in the conversation because swatches carry no series labels. Use no more than sixteen suggestions per call.\n3. Wait for its final choice, update only that series, and recheck the relevant background contrast. Flag duplicate colors and visibly close pairs without claiming a color-vision simulation or a quantitative distinction test that was not run.\n4. Continue through series needing a decision. Keep already accepted assignments stable; ask about any ambiguous Done message before applying it to a different series. Skip interactive selection for colors the user explicitly locks.\n5. For too many visually distinct categorical series, propose grouping, direct labels, small multiples, or additional patterns rather than promising that more hues solve the problem. Keep the final mapping understandable in grayscale through the chosen cues.\n\n## Deliver\n\nReturn the complete series-to-hex mapping, fixed background, marker/dash/pattern or direct-label cue for each series, and the exact contrast checks performed. Include a library configuration fragment only when the chart library and its API are known. Recommend checking the real rendered chart, especially adjacent marks and color-vision behavior. Do not claim the picker saved the mapping or previewed the chart.\n"
}

SHA-256 of public snapshot: 044bd2b0d07869e55f73c4da4468231001e0a03491568d3bf5cbb494eef08d75