Charts by Szum

Turn data into polished charts

After connecting, start a new chat and say: "Use Charts by Szum to create an editorial bar chart from this data: North 42, South 35, East 28, West 31.". Charts by Szum helps your AI turn plain-language requests and raw data into publication-ready charts.

What you can ask

create a polished editorial bar chart showing quarterly revenue of $1.2M, $1.8M, $2.4M, and $2.1M, with direct value labels.
Example path

MCP connector

Server URL: https://szum.io/mcp
Login: OAuth
Protocol: MCP 2025-11-25
Instructions: Chart authoring:
Every field-bearing semantic role names a data key: category/values for column, bar, line, area, and pie; x/y for scatter; category/values/stepKind for waterfall. Count values intentionally omit a value field. Series and encoding fields must also exist in the same rows.
Choose one chart type for the question: column/bar for categorical comparison, line/area for ordered trends, scatter for relationships, pie sparingly for part-to-whole, and waterfall for an ordered balance bridge.
A waterfall has one measure and ordered stepKind values: optional first start, one or more signed changes, optional subtotals, and a final total. Subtotal and total amounts are null; the compiler calculates their balances. Do not add series or aggregation. Value labels default outside bars, and focus: "changes" shows large balances as points.
Use long values with an optional series field, wide values with explicit fields, or count values. Aggregation belongs in the value role.
Column and bar support grouped or stacked multi-series arrangements. Area charts stack by default and support normalization.
Prefer native labels and references; use annotations for authored explanatory text.
Axes, color/size legends, and line points accept true for default settings, false to hide, or an options object that enables them. Omission keeps defaults. Stack totals use a boolean. Do not add display to these objects.

Tools

Reads data9

Changes data3

Deletes or can't be undone2

Skills

None in the store listings or on the server

What you can ask

create a polished editorial bar chart showing quarterly revenue of $1.2M, $1.8M, $2.4M, and $2.1M, with direct value labels.

Example path

MCP connector

Server URL

https://szum.io/mcp

OAuth

Protocol

MCP 2025-11-25

Instructions

Chart authoring:

Every field-bearing semantic role names a data key: category/values for column, bar, line, area, and pie; x/y for scatter; category/values/stepKind for waterfall. Count values intentionally omit a value field. Series and encoding fields must also exist in the same rows.

Choose one chart type for the question: column/bar for categorical comparison, line/area for ordered trends, scatter for relationships, pie sparingly for part-to-whole, and waterfall for an ordered balance bridge.

A waterfall has one measure and ordered stepKind values: optional first start, one or more signed changes, optional subtotals, and a final total. Subtotal and total amounts are null; the compiler calculates their balances. Do not add series or aggregation. Value labels default outside bars, and focus: "changes" shows large balances as points.

Use long values with an optional series field, wide values with explicit fields, or count values. Aggregation belongs in the value role.

Column and bar support grouped or stacked multi-series arrangements. Area charts stack by default and support normalization.

Prefer native labels and references; use annotations for authored explanatory text.

Axes, color/size legends, and line points accept true for default settings, false to hide, or an options object that enables them. Omission keeps defaults. Stack totals use a boolean. Do not add display to these objects.