Misata

Proof-backed synthetic data — realistic multi-table datasets with validation reports, from a sentence, YAML, or your own database.

Misata generates consistent, referentially-intact multi-table datasets from a plain-English description, a YAML schema file, or an existing database schema. Every normal generation run can also write an Oracle report: a shareable proof bundle for row counts, referential integrity, constraints, temporal consistency, locale/domain fit, privacy notes, fidelity scores, and reproducibility metadata.

No machine-learning model is required. No real data is needed.

Built for:

• Database seeding — fill dev and staging environments with production-like data
• Integration tests — relational fixtures with FK integrity across every table
• Demos and prototypes — realistic numbers, names, and distributions, no PII
• BI and dashboard development — data shaped like your real domain before launch

Install

pip install misata

Optional extras:

pip install "misata[llm]"        # multi-provider LLM schema generation
pip install "misata[documents]"  # PDF output via weasyprint
pip install "misata[advanced]"   # SDV/CTGAN statistical synthesis
pip install "misata[mcp]"        # MCP server — expose Misata to Claude, Cursor, and other AI agents

Use Misata from Claude / Cursor / Windsurf (MCP)

Misata ships a built-in Model Context Protocol server. Once configured, any MCP-compatible AI assistant can generate realistic synthetic data for you from natural language — no Python required on your end.

1. Install:

pip install "misata[mcp]"

2. Add to Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "misata": {
      "command": "misata-mcp"
    }
  }
}

Restart Claude Desktop. Then just ask:

"Generate a fintech dataset with 1 000 customers, payments, and a 2% fraud rate."

"Show me what tables Misata would produce for an HR system with 200 employees."

"I need SaaS data: MRR from $50k in January, doubled by December, with a Q3 slump."

Claude calls Misata, writes CSVs to disk, and returns the file paths plus a preview of each table. See the MCP guide for Cursor/Windsurf/Zed setup and all five available tools.

Quick start

misata generate \
  --story "Brazilian fintech with R$ payments, CPF verification, and 3% fraud" \
  --rows 1000 \
  --output-dir ./demo_data

# Writes CSVs plus:
# ./demo_data/oracle_report.json

import misata

# One sentence → multi-table DataFrame dict
tables = misata.generate("A SaaS company with 5k users, monthly subscriptions, and 20% churn")

print(tables["users"].head())
print(tables["subscriptions"].head())

# Or from the CLI
misata generate --story "A SaaS company with 5k users and 20% churn" --rows 5000

Misata Oracle

The Oracle report is Misata's proof layer. It separates hard guarantees from advisory realism checks so generated

... View full README on GitHub