Generate Synthetic Dataset

Overview

You declare the columns you want and a seed string; we return that many rows of invented values. Twelve generators cover integers, decimals, booleans, strings, categories, dates, date-times and UUID-shaped identifiers, and each cell is derived independently from the digest of your seed, the column name, the row index and the generator — so the same request always returns the same table, and adding a column or raising the row count leaves every cell you already had byte-identical. There is no embedded corpus of real names, addresses or companies here, no value carries a valid checksum, and every email domain must be one of the reserved names that cannot reach a real mailbox. Bounds for dates come from you, because this skill reads no clock.

Test data that looks real should not be real, and a fixture nobody can regenerate is a file somebody has to keep.

When to use it

Scope

Non-goals

• Does not embed any corpus of real names, addresses, companies or identifiers
• Does not produce a checksum-valid IBAN, VAT number, card number or national identifier
• Does not pseudonymise records you already hold — data.anonymize_dataset does that
• Does not guarantee any generated value is unique or unused in the real world
• Does not read a clock, so every date range is a caller input

Properties

Capabilities

Cost & billing

Flat rate per invocation

File handling

How we handle your data

Input schema

Output schema

API contract

curl -sS -X POST "$BASE/v1/execute" -H "Authorization: Bearer $KEY" -H 'Content-Type: application/json' -d '{"skillKey":"data.generate_synthetic_dataset","input":{"schema":{"columns":[{"name":"id","type":"INTEGER","generator":"SEQUENTIAL_INTEGER","params":{"start":1,"step":1}}]},"count":2,"seed":"fixture-a"}}'

import { Squillify } from "@squillify/sdk";

const sq = new Squillify({ apiKey: process.env.SQUILLIFY_API_KEY });

const { output } = await sq.execute("data.generate_synthetic_dataset", {"schema":{"columns":[{"generator":"SEQUENTIAL_INTEGER","params":{"start":1,"step":1},"type":"INTEGER","name":"id"},{"generator":"CATEGORICAL","params":{"values":["free","pro"]},"type":"ENUM","name":"tier"}]},"seed":"fixture-a","count":2});

Errors

Other limits

Limitations

• Does not embed any corpus of real names, addresses, companies or identifiers
• Does not produce a checksum-valid IBAN, VAT number, card number or national identifier
• Does not pseudonymise records you already hold — data.anonymize_dataset does that
• Does not guarantee any generated value is unique or unused in the real world
• Does not read a clock, so every date range is a caller input
• Generating tokens, nonces or any security identifier
• Producing values for an identity-bearing production field
• Anonymising a real dataset

How to call

POST /v1/execute (requires authentication)

{
  "skillKey": "data.generate_synthetic_dataset",
  "input": {
    "schema": {
      "columns": [
        {
          "generator": "SEQUENTIAL_INTEGER",
          "params": {
            "start": 1,
            "step": 1
          },
          "type": "INTEGER",
          "name": "id"
        },
        {
          "generator": "CATEGORICAL",
          "params": {
            "values": [
              "free",
              "pro"
            ]
          },
          "type": "ENUM",
          "name": "tier"
        }
      ]
    },
    "seed": "fixture-a",
    "count": 2
  }
}

Examples

{
  "schema": {
    "columns": [
      {
        "generator": "SEQUENTIAL_INTEGER",
        "params": {
          "start": 1,
          "step": 1
        },
        "type": "INTEGER",
        "name": "id"
      },
      {
        "generator": "CATEGORICAL",
        "params": {
          "values": [
            "free",
            "pro"
          ]
        },
        "type": "ENUM",
        "name": "tier"
      }
    ]
  },
  "seed": "fixture-a",
  "count": 2
}

{
  "warnings": [],
  "_meta": {
    "generatorProfile": "synthetic-1",
    "seedEcho": "fixture-a"
  },
  "table": {
    "rows": [
      [
        1,
        "free"
      ],
      [
        2,
        "pro"
      ]
    ],
    "columns": [
      "id",
      "tier"
    ]
  },
  "work": {
    "projected": 4,
    "budget": 2000000
  },
  "summary": {
    "rowCount": 2,
    "columnCount": 2,
    "cells": 4,
    "bytes": 9
  }
}

{
  "count": 1,
  "schema": {
    "columns": [
      {
        "generator": "SEQUENTIAL_INTEGER",
        "params": {
          "start": 10,
          "step": 1
        },
        "type": "INTEGER",
        "name": "id"
      }
    ]
  },
  "seed": "fixture-rows",
  "output": "ROWS"
}

{
  "warnings": [],
  "_meta": {
    "generatorProfile": "synthetic-1",
    "seedEcho": "fixture-rows"
  },
  "table": {
    "rows": [
      {
        "id": 10
      }
    ]
  },
  "work": {
    "projected": 1,
    "budget": 2000000
  },
  "summary": {
    "rowCount": 1,
    "columnCount": 1,
    "cells": 1,
    "bytes": 2
  }
}

Recommended use cases

• Filling a test database with realistic-shaped rows
• Building a reproducible fixture for a load test
• Demonstrating a dashboard without exposing customer data

Avoid for

• Generating tokens, nonces or any security identifier
• Producing values for an identity-bearing production field
• Anonymising a real dataset

Related skills

Composable with

Safety & privacy

• Runs without calling any external service.
• Your payloads are never stored permanently.
• Telemetry captures metadata only — never your payload contents.
• We never train models on your payloads.
• No human reviews your payloads.

Tags