Agent Paranoid Android

Safety-first, deterministic synthetic test data generation from CSV structure, safe profiles, reviewed DatasetSpec files, and allowlisted Trino metadata. Source rows are profiled, never shuffled or copied into generated output.

Read the documentation for tutorials, concepts, configuration, MCP setup, and troubleshooting.

Current package version: 0.8.1. The package is agent-paranoid-android; the CLI is test-data-agent.

Install

Python 3.11 or newer is required.

python3 -m pip install agent-paranoid-android
test-data-agent doctor

Explore the CLI without needing to remember a command:

test-data-agent
test-data-agent examples
test-data-agent generate --help

Install only features you use:

python3 -m pip install "agent-paranoid-android[parquet]"
python3 -m pip install "agent-paranoid-android[mcp,trino]"
python3 -m pip install "agent-paranoid-android[all]"

Quickstart

Clone the repository to use its safe fictional fixture:

git clone https://github.com/wa-pis/agent-paranoid-android.git
cd agent-paranoid-android

test-data-agent generate-from-csv tests/fixtures/customers.csv \
  --count 25 \
  --seed 12345 \
  --format csv \
  --output out/customers.csv

A successful run reports:

Generated synthetic dataset: out | rows: customers=25 | seed: 12345 | validation: passed | source rows copied: no

The output folder contains the generated data plus:

• csv_profile.json;
• dataset_spec.json;
• validation_report.json;
• generation_manifest.json.

Review the manifest and effective spec before accepting a new dataset.

Choose A Guide

Safety

The project derives bounded metadata such as field types, null ratios, ranges, masked patterns, and safe low-cardinality distributions. It rejects or bounds:

• raw detected PII, credentials, tokens, and private keys in profiles;
• source-row copying and source/output path reuse;
• path traversal and symlink escapes through generator MCP tools;
• unrestricted SQL and write operations through Trino tools;
• oversized input, output, rule, query, and generation work.

Human review is still required for ambiguous identifiers, rare free text, inferred relationships, and organization-specific privacy policy.

Development

python3 -m pip install "uv==0.11.23"
uv sync --frozen --all-extras --no-install-project
uv sync --frozen --all-extras --no-editable --no-build-isolation
uv run --no-sync scripts/check_release.sh

See Contributing, Security Policy, and Changelog.

Releases use tokenless PyPI Trusted Publishing and include verified wheel and source distributions, checksums, SBOMs, and GitHub attestations.

AI-Assisted Development

This project is developed with substantial assistance from AI coding tools. All changes remain subject to human review, automated testing, and the same security requirements as manually written code. Do not send production data, raw PII, credentials, or tokens to an AI system while working with the project.

The name is a nod to Radiohead's "Paranoid Android". This project is unaffiliated with Radiohead or related rights holders.

License

MIT