42-True is offered in the spirit of open science. At this stage the repository is a concept — a schema and a set of recipes for a corpus that does not yet exist — so the most valuable contributions are conceptual and structural.
- Critique of the schema. Does the intent pair
( Id, C, M, O )capture the right structure? What's missing, what's over-specified? - Classification & taxonomy. Proposals for the machine-addressable schema and the three-tier resolution architecture (paper §V).
- Privacy architecture. Scrutiny of the zero-knowledge boundary and the unlinkability guarantees (paper §VI).
- Production network. Ideas on how the corpus could actually be produced — declaration surface, market design, outcome attestation (paper §VIII).
- Stewardship. Models for holding the result as a commons (paper §VII).
This repository must never contain real declared-intent data. Any example
records committed here must be synthetic and clearly labelled as such (see
data/example_intent_pairs.jsonl). The whole
premise is that genuine declarations are gathered only under a consented,
unobservable protocol — not pasted into a public repo.
pip install -e ".[dev]"
pytest
ruff check .Models are frozen (immutable) by design: prefer constructing new records over mutating existing ones, and keep that invariant in any new code.
Open an issue using one of the issue forms — there is
one per contribution type above (schema, classification, privacy, production
network, stewardship, and prior-work/corrections). For private or sensitive
matters, reach the author at shawn@profila.com.