# Related Standards

> How Truth Machine relates to provenance, normative language, schemas, canonicalization, and AI discovery conventions.

Truth Machine composes responsibilities that existing standards address in
part. It does not redefine their wire formats or claim to replace them.

## W3C PROV

[PROV-DM](https://www.w3.org/TR/prov-dm/) defines a domain-agnostic model for
entities, activities, agents, derivation, attribution, and responsibility.
Truth Machine evidence and change ancestry can be mapped to those concepts.

PROV does not by itself choose bounded source authority, identify current
truth, bind exact human review, or define the atomic advancement boundary in
this specification. An implementation may export PROV without using RDF or
PROV as its internal schema.

## Normative language

Requirement keywords follow
[BCP 14](https://www.rfc-editor.org/rfc/rfc8174.html), combining RFC 2119 and
RFC 8174. They are normative only when capitalized and attached to a stable
`TM-*` requirement identifier.

## JSON Schema

The reference packet schemas use
[JSON Schema Draft 2020-12](https://json-schema.org/draft/2020-12). They
describe one implementation profile, not the universal semantic model.
Conformance validators must still enforce cross-record and transition
invariants that a single document schema cannot express.

## Canonicalization and identity

[RFC 8785](https://www.rfc-editor.org/rfc/rfc8785.html) defines JSON
Canonicalization Scheme for repeatable hashing and signing. An implementation
that reviews JSON by digest may use JCS. Truth Machine requires exact review
identity but does not require JSON or JCS.

The Git-native profile uses Git object and commit identities plus explicit
SHA-256 file manifests where byte identity outside Git must remain portable.

## AI-readable documentation

[`llms.txt`](https://llmstxt.org/) is an emerging convention for a concise
Markdown site map aimed at inference-time use. This project publishes it as a
convenience, raw page-level Markdown, a compact core context, a full context,
and a structured manifest. None replaces canonical documentation or ordinary
web standards such as semantic HTML, `robots.txt`, and sitemaps.

The publication test suite treats every generated form as derived output and
fails on drift.