# HIPG Formal Toy v0.4

Run:

```bash
python hipg_toy_v0_4.py
python -m unittest -v test_hipg_v0_4.py
```

Key v0.4 additions:

- binary-reward active causal experiment selection;
- Bayesian quotient-equivalence posterior;
- OOD surface-probe permutation inference;
- counterfactual infeasibility relaxation analysis;
- JSON-Schema certificate validation + hash lineage;
- restricted typed permission/scope contract DSL.

All v0.3 canonical cases remain in `benchmark_cases.jsonl`.