Documentation

Tools Reference

Data & Analytics

Four tools for parsing, querying, analyzing, and visualizing data. The most powerful category for data-focused agents.

csv_parse Standard

Parse CSV (Comma-Separated Values) data into a structured array of objects.

TypeScript
tools.csvParse

Parameters

ParameterTypeRequiredDescription
csvDatastring✓Raw CSV string (with header row)
delimiterstring—Column delimiter (default: ,)

Returns

TypeScript
{
  headers: ["name", "revenue", "region"],
  rows: [
    { name: "Alpha", revenue: "50000", region: "US" },
    { name: "Beta", revenue: "75000", region: "EU" },
  ],
  rowCount: 2,
  columnCount: 3
}

data_query Standard

Query datasets using operations like filtering, sorting, grouping, and aggregation.

TypeScript
tools.dataQuery

Parameters

ParameterTypeRequiredDescription
dataobject[]✓Array of data objects to query
operationstring✓One of: filter, sort, group, aggregate, select, limit
paramsobject✓Operation-specific parameters (see below)

Operations

  • filter — Filter rows by a condition: { field, operator, value }
  • sort — Sort by field: { field, direction }
  • group — Group by field: { field }
  • aggregate — Aggregate values: { field, function } (sum, avg, min, max, count)
  • select — Select specific fields: { fields: string[] }
  • limit — Limit results: { count }

statistics Standard

Compute statistical metrics on numeric data arrays.

TypeScript
tools.statistics

Parameters

ParameterTypeRequiredDescription
datanumber[]✓Array of numeric values
metricsstring[]—Which metrics to compute (default: all)

Available Metrics

meanmedianmodeminmaxrangesumcountvariancestddevpercentile25percentile75skewnesskurtosis

Returns

TypeScript
{
  mean: 51666.67,
  median: 50000,
  mode: null,
  min: 30000,
  max: 75000,
  range: 45000,
  sum: 155000,
  count: 3,
  variance: 338888888.89,
  stddev: 18409.73,
  percentile25: 40000,
  percentile75: 62500,
  skewness: 0.12,
  kurtosis: -1.5
}

generate_chart Standard

Generate data visualizations using the Vega-Lite specification format.

TypeScript
tools.generateChart

Parameters

ParameterTypeRequiredDescription
chartTypestring✓bar, line, pie, scatter, or area
dataobject[]✓Array of data points
xFieldstring✓Data field for x-axis
yFieldstring✓Data field for y-axis
titlestring—Chart title

Chart Types

  • bar — Categorical comparisons (revenue by region)
  • line — Trends over time (monthly sales)
  • pie — Proportional distribution (market share)
  • scatter — Correlations (price vs. quality)
  • area — Cumulative trends (growth over time)

Returns

TypeScript
{
  spec: { /* Vega-Lite JSON specification */ },
  imageUrl: "data:image/png;base64,..."
}

Example: Full Data Pipeline

A typical data agent chains these four tools:

  1. csv_parse → Parse uploaded CSV
  2. data_query → Filter and aggregate
  3. statistics → Compute key metrics
  4. generate_chart → Visualize results

Next Steps