motleyai/update-chart

Update Chart

Create or modify chart blocks using the update_chart_block MCP tool. You provide a structured query and chart configuration directly — no LLM intermediary.

How It Works

1. Call update_chart_block with a query (what data to fetch) and chart_details (how to render it)
2. The query is validated and the chart configuration is saved to the block
3. Call render_chart to see the rendered PNG (this also triggers resolution)

Returns: Confirmation that the chart was updated with a data preview. Does NOT return a rendered image — use render_chart for that.

Chart Type Guidance

| Type | Best For | Example | |------|----------|---------| | BAR | Categorical comparisons, rankings, time series with few points | Revenue by region | | LINE | Trends over time, continuous data, multiple series comparison | Monthly active users over 12 months | | PIE | Part-to-whole relationships, distribution (use sparingly, max 5-7 segments) | Revenue distribution by category | | FUNNEL | Conversion stages, sequential process drop-off | Sales funnel from lead to close |

Tips

• Bar vs Line: Use bars for categorical comparisons or few time points. Use lines for trends with many time points.
• Pie charts: Only effective with 5-7 or fewer segments. For more categories, use a bar chart.
• Dual axis: Useful when comparing measures with different scales (e.g., revenue in dollars vs count of orders). Set y_axis: "right" on the secondary series.
• Stacked bars: Use the same x_axis dimension with multiple measures, all as BAR type.
• Horizontal bars: Use BAR type with dimensions on the y-axis — good for long category labels or ranking displays.

Examples

Time Series — Monthly Revenue

update_chart_block(
    location={doc_id: 42, slide_name: "Revenue Trends", block_name: "revenue_chart"},
    query={
        measures: [{name: "total_revenue", cube_name: "revenue"}],
        time_dimension: {
            dimension: {name: "created_at", cube_name: "revenue"},
            granularity: "month"
        },
        limit: 12,
        order: [{column: {name: "created_at", cube_name: "revenue"}, order: "ASC"}]
    },
    chart_details={
        series_default: {type: "LINE", y_axis: "left", number_format: {style: "currency"}, show_values: false},
        x_axis: {lines: false, label: false},
        y_axis: {lines: true, label: "Revenue"},
        y_right_axis: {lines: false},
        title: "Monthly Revenue"
    }
)

Categorical — Top Regions

update_chart_block(
    location={doc_id: 42, slide_name: "Regional Performance", block_name: "region_chart"},
    query={
        measures: [{name: "total_revenue", cube_name: "sales"}],
        dimensions: [{name: "region", cube_name: "sales"}],
        limit: 10,
        order: [{column: {name: "total_revenue", cube_name: "sales"}, order: "DESC"}]
    },
    chart_details={
        series_default: {type: "BAR", y_axis: "left", number_format: {style: "currency"}, show_values: true},
        x_axis: {lines: false, label: "Region"},
        y_axis: {lines: true, label: "Revenue"},
        y_right_axis: {lines: false},
        title: "Top 10 Regions by Revenue"
    }
)

Dual Axis — Revenue vs Count

update_chart_block(
    location={doc_id: 42, slide_name: "Overview", block_name: "dual_chart"},
    query={
        measures: [
            {name: "total_revenue", cube_name: "orders"},
            {name: "order_count", cube_name: "orders"}
        ],
        time_dimension: {
            dimension: {name: "created_at", cube_name: "orders"},
            granularity: "month"
        },
        limit: 12
    },
    chart_details={
        series: {
            "total_revenue": {type: "BAR", y_axis: "left", number_format: {style: "currency"}, show_values: false},
            "order_count": {type: "LINE", y_axis: "right", number_format: {style: "decimal"}, show_values: false}
        },
        series_default: {type: "BAR", y_axis: "left"},
        x_axis: {lines: false, label: false},
        y_axis: {lines: true, label: "Revenue"},
        y_right_axis: {lines: true, label: "Orders"},
        title: "Revenue vs Order Count"
    }
)

Pie — Distribution

update_chart_block(
    location={doc_id: 42, slide_name: "Breakdown", block_name: "pie_chart"},
    query={
        measures: [{name: "total_revenue", cube_name: "products"}],
        dimensions: [{name: "category", cube_name: "products"}],
        limit: 5,
        order: [{column: {name: "total_revenue", cube_name: "products"}, order: "DESC"}]
    },
    chart_details={
        series_default: {type: "PIE", y_axis: "left", number_format: {style: "currency"}, show_values: true},
        x_axis: {lines: false},
        y_axis: {lines: false},
        y_right_axis: {lines: false},
        title: "Revenue by Category",
        legend: {enabled: true, location: "auto", orientation: "VERTICAL"}
    }
)

Resolution Behavior

Charts are NOT resolved immediately after update_chart_block. The chart config is saved, but no data is fetched or rendered yet.

To see the chart:

• Individual chart: Call render_chart(location={doc_id: ..., slide_name: ..., block_name: ...}) — this triggers resolution and returns a PNG image

Verifying Charts

Always call render_chart after update_chart_block to visually verify the result.

render_chart(
    location={doc_id: 42, slide_name: "Revenue Trends", block_name: "revenue_chart"},
    width=800,
    height=600
)

Check the returned image for:

• Correct chart type
• Appropriate axis labels and scales
• Data looks reasonable (no obviously wrong values)
• Legend is readable
• Time series is in chronological order

If the chart doesn't look right, call update_chart_block again with adjusted query or chart_details.

Troubleshooting

| Issue | Solution | |-------|----------| | Wrong chart type | Set series_default.type explicitly (BAR, LINE, PIE, FUNNEL) | | Missing data points | Check that the model has data for the requested time range. Use inspect_model to verify. | | Too many categories | Add limit to the query | | Wrong time granularity | Set time_dimension.granularity explicitly | | Axis scale issues | Set scale: "LOG" on the axis, or use dual axis with y_axis: "right" | | Unknown measure/dimension | Use inspect_model to see exact names | | Wrong cube | Set cube_name on each measure/dimension explicitly |

Related Skills

• For exploring models before building charts: see the explore-model skill
• For understanding model schemas: see model-guide.md