Reasoning you can verify — by people and machines.
Big decisions — in policy, in markets, in AI — rest on reasoning you’re asked to trust but never get to see. OpenReason is an open standard that puts it in the open: the sources, the assumptions, the trade-offs, structured so anyone can check it. Since v0.3, a machine can verify it cryptographically before acting on it.
Someone just made a decision that affects you.
Do you know how they got there?
For people
Document, inspect, fork and challenge the reasoning behind a decision. Trust comes from argument and cited sources — not authority. This is the original civic mission.
How it works →For machines
Verify a dataset offline, enforce what may be done with it, and prove use — a signed passport an agent checks itself, with no new crypto. Added in v0.3.
The machine layer →Not a report. A simulatable object.
An ORP document is reasoning you can re-run — five structured layers, from where the data came from to how the conclusion was reached, that anyone can fork and re-run with different assumptions. Disagreement becomes locatable: you can point to the exact assumption two parties diverge on.
In v0.3 the object is sealed with a signature — so a machine can verify the whole thing offline and know exactly how far to trust it.
Data products
orp conform — claim it, gate CI.Write your first document
Apache-2.0 · Python · runs locally.
pip install orp-validator
orp new my-policy.yaml --compliance basic
orp validate my-policy.yaml