Getting started

Choose the result you need. Generating synthetic identities does not require learning the agent-authorisation workflow.

Generate connected synthetic identities

Use Python 3.12 or newer, preferably in a virtual environment:

python -m venv .venv
# Activate the environment using the command for your operating system.
pip install idcognito-synthworld==0.18.0
synthworld generate --seed 20260719 --persona-count 10 --output world.json

On macOS/Linux, activate with source .venv/bin/activate. On Windows PowerShell, use .venv\Scripts\Activate.ps1 before running pip.

You now have ten fictional personas and their connected relationships in world.json. Continue with identity worlds to inspect the data and check reproducibility. For matching or privacy evaluation, use identity resolution or privacy and exposure.

Test agent authorisation

Start with the complete demo. It provides a browser preview and runs generation, teaching policies, and scoring without an unwritten adapter or a pre-existing trace file. The packaged synthworld-demo command is included in 0.18.0; the guide covers installation and source-checkout usage.

After the demo, follow agent authority to understand the trace contract, then development pipelines for integration. Connecting your real system is separate from running the demo.

Develop or run repository examples

From a repository checkout:

uv sync --locked --all-groups
uv run python examples/evaluate_all.py --predictions-dir predictions

That walkthrough demonstrates five foundational public-input adapters, including extraction and identity resolution. It is not an exhaustive evaluator inventory.

Reference pages describe current main unless they state a released version. Pin the package version used by your integration. The legacy user guide remains a compatibility index for historical links.