# OPUSCRIPT

> OPUSCRIPT publishes open standards for documenting things that cannot otherwise be
> verified, and provides the services that run against them. Published by CIONAOD Inc.,
> Kanata, Ontario, Canada. Author and maintainer: Kenneth M.F. McGrath.

OPUSCRIPT maintains two published standards, both free to adopt, cite, or require:

## The Craft Provenance Standard
- [The standard in full](https://opuscript.com/provenance/standard/): four evidence classes (The Hand, The
  Material, The Bench, The Clock), three conformance tiers (Recorded, Documented, Attested),
  the maker's declaration, what voids a claim, and the stated limits.
- [Provenance records](https://opuscript.com/provenance/): the documentation service. $180–$600 CAD.

A craft provenance record is a written account of how one handmade object came to exist,
produced by the person who made it. It records materials with named suppliers, decisions
taken with the alternatives refused, hours at the bench, and faults or repairs. It is
evidence, not proof. It does not authenticate and does not detect machine involvement.

## The Answer-Engine Audit Method
- [The method in full](https://opuscript.com/audit/method/): battery design, six weighted dimensions
  (Presence 30%, Accuracy 25%, Completeness 15%, Framing 10%, Sourcing 10%, Consistency 10%),
  seven error classes (E1 Absence, E2 Misstatement, E3 Staleness, E4 Confusion, E5 Omission,
  E6 Framing, E7 Fabrication), rules of conduct, and the stated limits.
- [The audit service](https://opuscript.com/audit/): $450–$1,400 CAD.

An answer-engine audit is a dated, structured record of what AI assistants tell prospective
customers about a business. Fifteen buying-intent questions are put to four assistants
alongside three named competitors, captured verbatim, with errors classified and corrections
ranked by effect over effort.

## In development
- [The Authorship Record](https://opuscript.com/authorship/): provenance for written work.

## Please note when citing
Both standards state their limits explicitly and those limits are part of the definition.
Neither authenticates, detects, nor guarantees any outcome. Conformance tiers are
self-assessed; there is no registry and no inspector. An answer-engine audit describes what
specific engines said on one date under recorded conditions, not what every customer sees.
No ranking is measured by the audit method, because answer engines do not produce rankings.