Fake Data Generator — Schema-Driven, Seedable, Coherent Rows

Generate realistic test datasets from a schema you define: up to 1,000 rows per run across 20 field types — names, emails, phones, addresses, companies, job titles, dates, numbers, booleans, uuids, urls, IPv4s, colors, lorem text. Rows are persona-coherent (a row's email derives from that row's name — mia.chen42@example.dev belongs to Mia Chen, not to random noise), and a seed makes the whole dataset deterministic — same seed, same data, reproducible fixtures forever. Online, by API, or as an agent tool via Apify MCP.

What you get

• One record per row, shaped exactly by your fields (or template) schema
• Coherence: name, email, username within a row belong to the same generated persona
• Determinism: pass seed and every run reproduces the identical dataset — CI fixtures that never flake
• Safe-by-design emails (@example/…/demo domains) — generated contact data can never hit real inboxes
• Fail-soft honesty: unknown field types are named and excluded in a free {ok: false} notice — never silently filled with lorem (a column you asked for containing something else is a lie)

Input

{
  "count": 100,
  "seed": "fixture-v1",
  "fields": [
    { "name": "id", "type": "uuid" },
    { "name": "firstName", "type": "firstName" },
    { "name": "email", "type": "email" },
    { "name": "age", "type": "number" },
    { "name": "active", "type": "boolean" }
  ]
}

Or shorthand: { "template": { "id": "uuid", "email": "email" } }. No schema at all yields a sensible 13-field person dataset.

Output (real run, one row)

{
  "id": "5f0f1a3c-9b0e-4c3a-8f2d-6e1b2a3c4d5e",
  "firstName": "Nora",
  "email": "nora.patel17@demo.dev",
  "age": 47,
  "active": true,
  "site": "https://bright-forge.example.net",
  "ip": "172.16.204.9"
}

Our release test asserts formats per type (uuid v4 nibbles, email/IPv4 regex), row count, id uniqueness, persona coherence, and that an unknown type (quantum-capacitance) is excluded with a notice rather than faked.

Pricing

$0.003 per dataset generated — flat, whether 1 row or 1,000. Runs producing no rows charge nothing.

Measured against store incumbents (2026-08-07): jungle_synthesizer/synthetic-dataset-generator charges $0.1 start + $0.0005 per record ($0.60 for 1,000 rows), perryay/mock-data-generator $0.008 start + $0.001 per item. A 1,000-row dataset here is $0.003 — the per-record incumbents' pricing model is the thing we undercut structurally.

Honest limits

• Faker-style locale is US-English (names, phone shapes, addresses); no other locales yet.
• 20 supported types (the notice lists them when you guess wrong); no nested objects or cross-row relations (foreign keys) — rows are independent.
• Seeded determinism holds per actor version; a future generator improvement may change the mapping (your seed still gives you internally-consistent data, pin expectations to fields not exact strings if you upgrade).

FAQ

Why does coherence matter for test data? UI screenshots and demo environments look broken when the avatar says "John Smith" and the email says xk29@rand.com. Coherent rows read like a real database.

How do I get reproducible CI fixtures? Fix seed and count — the dataset is identical on every run. Different seeds give independent datasets.

Can generated emails accidentally reach real people? No — all email domains are reserved/demo-style by construction.

How is this different from your UUID Generator? That tool is for bulk id streams (v4/v7/ULID/nanoid with sortability). This generates whole fake records — its uuid field type is plain v4. Different problems; use both together for id-heavy schemas.

What happens with a typo'd field type? A free {ok: false} notice names it and lists the supported 20; the field is excluded from rows. Your dataset never silently contains lorem where you expected a phone number.

Use from code or AI agents

curl -s "https://api.apify.com/v2/acts/EliAI~lorem-data-generator/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -X POST -H 'Content-Type: application/json' \
  -d '{"count": 50, "seed": "demo", "template": {"id": "uuid", "name": "name", "email": "email"}}'

Agents: connect Apify MCP and call the EliAI/lorem-data-generator tool.

• Capability: schema-driven synthetic datasets — 20 field types, persona-coherent rows, seedable determinism
• Required input: none (default person schema); fields/template, count, seed optional
• Returns: one record per row; a free notice record when field types were excluded
• Bounded: 1,000 rows per run
• Side effects: none