Skip to main content
Glama

suggestChart

Recommend appropriate chart types for data visualization using Semiotic library. Provide sample data to get ranked suggestions with example properties.

Instructions

Recommend Semiotic chart types for a given data sample. Pass { data: [...] } with 1-5 sample objects. Optionally pass intent to narrow suggestions. Returns ranked recommendations with example props.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
dataYes1-5 sample data objects
intentNoVisualization intent to narrow suggestions

Implementation Reference

  • The suggestChartHandler function analyzes the provided data and returns recommended chart types, including confidence levels and sample props.
    async function suggestChartHandler(args: { data?: any[]; intent?: string }): Promise<ToolResult> {
      const data = args.data
      const intent = args.intent
    
      if (!data || !Array.isArray(data) || data.length === 0) {
        return {
          content: [{ type: "text" as const, text: "Pass { data: [{ ... }, ...] } with 1-5 sample data objects. Optionally include intent: 'comparison' | 'trend' | 'distribution' | 'relationship' | 'composition' | 'geographic' | 'network' | 'hierarchy'." }],
          isError: true,
        }
      }
    
      const sample = data[0]
      if (!sample || typeof sample !== "object") {
        return {
          content: [{ type: "text" as const, text: "Data items must be objects with key-value pairs." }],
          isError: true,
        }
      }
    
      const keys = Object.keys(sample)
      const suggestions: Array<{ component: string; confidence: string; reason: string; props: Record<string, string> }> = []
    
      // Classify fields
      const numericFields: string[] = []
      const stringFields: string[] = []
      const dateFields: string[] = []
      const geoFields: { lat?: string; lon?: string } = {}
      const networkFields: { source?: string; target?: string; value?: string } = {}
      const hierarchyFields: { children?: string; parent?: string } = {}
    
      for (const key of keys) {
        const values = data.map(d => d[key]).filter(v => v != null)
        if (values.length === 0) continue
    
        const first = values[0]
        if (typeof first === "number") {
          numericFields.push(key)
        } else if (typeof first === "string") {
          if (/^\d{4}[-/]\d{2}/.test(first) && !isNaN(Date.parse(first))) {
            dateFields.push(key)
          } else {
            stringFields.push(key)
          }
        }
    
        const kl = key.toLowerCase()
        if (kl === "lat" || kl === "latitude") geoFields.lat = key
        if (kl === "lon" || kl === "lng" || kl === "longitude") geoFields.lon = key
        if (kl === "source" || kl === "from") networkFields.source = key
        if (kl === "target" || kl === "to") networkFields.target = key
        if (kl === "value" || kl === "weight" || kl === "amount") networkFields.value = key
        if (kl === "children" || kl === "values") hierarchyFields.children = key
        if (kl === "parent") hierarchyFields.parent = key
      }
    
      const hasTime = dateFields.length > 0
      const hasCat = stringFields.length > 0
      const hasNum = numericFields.length > 0
      const hasGeo = geoFields.lat && geoFields.lon
      const hasNetwork = networkFields.source && networkFields.target
      const hasHierarchy = hierarchyFields.children || hierarchyFields.parent
    
      // Network data
      if (hasNetwork && (!intent || intent === "network")) {
        const src = networkFields.source!
        const tgt = networkFields.target!
        if (networkFields.value) {
          suggestions.push({
            component: "SankeyDiagram",
            confidence: "high",
            reason: `Data has ${src}→${tgt} with ${networkFields.value} — ideal for flow visualization`,
            props: { edges: "data", sourceAccessor: `"${src}"`, targetAccessor: `"${tgt}"`, valueAccessor: `"${networkFields.value}"` },
          })
        }
        suggestions.push({
          component: "ForceDirectedGraph",
          confidence: networkFields.value ? "medium" : "high",
          reason: `Data has ${src}→${tgt} edges — force layout shows network structure. Nodes are auto-inferred from edges when not provided.`,
          props: { edges: "data", sourceAccessor: `"${src}"`, targetAccessor: `"${tgt}"` },
        })
      }
    
      // Hierarchy data
      if (hasHierarchy && (!intent || intent === "hierarchy")) {
        suggestions.push({
          component: "Treemap",
          confidence: "high",
          reason: `Data has nested ${hierarchyFields.children || "parent"} structure — treemap shows hierarchical proportions`,
          props: { data: "rootObject", childrenAccessor: `"${hierarchyFields.children || "children"}"`, ...(numericFields[0] ? { valueAccessor: `"${numericFields[0]}"` } : {}) },
        })
        suggestions.push({
          component: "TreeDiagram",
          confidence: "medium",
          reason: "Tree layout shows hierarchical relationships",
          props: { data: "rootObject", childrenAccessor: `"${hierarchyFields.children || "children"}"` },
        })
      }
    
      // Geographic data
      if (hasGeo && (!intent || intent === "geographic")) {
        const sizeField = numericFields.find(f => f !== geoFields.lat && f !== geoFields.lon)
        suggestions.push({
          component: "ProportionalSymbolMap",
          confidence: "high",
          reason: `Data has ${geoFields.lat}/${geoFields.lon} coordinates — map shows spatial distribution`,
          props: { points: "data", xAccessor: `"${geoFields.lon}"`, yAccessor: `"${geoFields.lat}"`, ...(sizeField ? { sizeBy: `"${sizeField}"` } : {}) },
        })
      }
    
      // Time series
      if (hasTime && hasNum && (!intent || intent === "trend")) {
        const timeField = dateFields[0]
        const valueField = numericFields[0]
        suggestions.push({
          component: "LineChart",
          confidence: "high",
          reason: `Data has dates (${timeField}) and numeric values (${valueField}) — line chart shows trends over time`,
          props: { data: "data", xAccessor: `"${timeField}"`, yAccessor: `"${valueField}"`, ...(hasCat ? { lineBy: `"${stringFields[0]}"`, colorBy: `"${stringFields[0]}"` } : {}) },
        })
        if (hasCat) {
          suggestions.push({
            component: "StackedAreaChart",
            confidence: "medium",
            reason: `Multiple categories (${stringFields[0]}) over time — stacked area shows composition trends`,
            props: { data: "data", xAccessor: `"${timeField}"`, yAccessor: `"${valueField}"`, areaBy: `"${stringFields[0]}"`, colorBy: `"${stringFields[0]}"` },
          })
        }
  • The suggestChart tool is registered using srv.tool, defining the input schema and connecting it to the suggestChartHandler.
    srv.tool(
      "suggestChart",
      "Recommend Semiotic chart types for a given data sample. Pass { data: [...] } with 1-5 sample objects. Optionally pass intent to narrow suggestions. Returns ranked recommendations with example props.",
      {
        data: z.array(z.record(z.string(), z.unknown())).min(1).max(5).describe("1-5 sample data objects"),
        intent: z.enum(["comparison", "trend", "distribution", "relationship", "composition", "geographic", "network", "hierarchy"]).optional().describe("Visualization intent to narrow suggestions"),
      },
      suggestChartHandler
    )

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/nteract/semiotic'

If you have feedback or need assistance with the MCP directory API, please join our Discord server