Kinra Corpus

Assessment identity

• Repository: kinra-ai/kinra-corpus
• Revision: 39fc3f1dfdf3f9690d02cf2023506989ecf4d501
• Assessment date: 2026-08-09
• Evidence class: bounded maintainer attestation; organizational evidence remains private

Domain

Kinra Corpus governs stable organization-level truth: foundation, source authority, operating model, and later organizational domains as they are actually required.

Mechanisms observed

• untrusted arrivals, evidence, reviewed records, changes, and current selection remain semantically distinct;
• evidence packets preserve provenance, authority scope, files, and SHA-256 identities;
• accepted records are immutable and later superseded;
• a typed current.toml selects exactly one accepted record per domain;
• every change names its expected canonical revision, exact domain transitions, evidence, actor, reviewer, rationale, and uncertainty;
• accepted records, current selection, and change land in one commit;
• the default reader uses canonical origin/main, not the worktree;
• proposal validation is a separate explicit mode; and
• the accepted bootstrap mechanically reads MATCH without any facet or publication surface.

What it paid to teach

Kinra Corpus is the decisive second full implementation of the portable evidence-to-truth grammar. It applies the mechanism to organizational domains rather than a manufacturing account and deliberately begins with only the domains reality required.

Its explicit absence of a facet proves that facets are not part of the truth-machine core. That observation supplied the final evidence for creating this successor repository.

Limits

The schema has one human reviewer and a Git-native atomic boundary. Its record types, directory layout, domain names, and authority policy remain one implementation profile. The initial accepted transition proves structural operation, not the truth of every later organizational claim.