Section 117 of 440
APPENDIX AB: USING THE COCKPIT AND ENGINE
HCE record governance and version control
The Cockpit and Engine are the working implementation of HCE’s record architecture. Every claim-critical record must identify the constitutional and technical meanings under which it was created. Revision 7N preserves the Revision 7L RC2 technical baseline: HCE constitutional semantics v1.0, Constitution editorial release v1.0.2, Capability Genome Schema v1.0 and Semantic Validator v1.0. The Lab consolidation adds research-governance identity only; it does not claim that the Cockpit or Engine implements the Lab’s complete archive or promotion authority.
Control | Required record |
|---|---|
Version provenance | HCE constitution, Genome schema, semantic validator, registry, module, protocol and criterion versions. |
Structural validation | The record conforms to the declared JSON Schema; required fields and value types are present. |
Semantic validation | Cross-field invariants are checked, including safety consistency, maturity evidence, lifecycle logic and actor/configuration boundaries. |
Assessment provenance | Declaration ID and digest, task source, evidence mode, support, data coverage, assessor or instrument details and immutable event linkage. |
Migration | A declared transformation with source and target versions, test results, exceptions and a readable legacy path. |
Round-trip integrity | Canonical records can be imported and exported without silent semantic or hash-changing mutation. |
Rollback | The pre-integration registry and readers remain recoverable until the migrated release passes every gate. |
Privacy-preserving history | Append-only evidence may use separated identity keys, redaction, tombstoning or lawful erasure while preserving the minimum governance trace. |
The Engine must never compute an overall capability, maturity, viability or human-worth score from these fields. It may filter records by burden, equipment, accessibility, evidence maturity, maintenance or transfer, but those views remain decision aids rather than rankings of people.
Where technology is part of the tested system, the record should preserve human-only, assisted and combined baselines when safe and relevant. It should also name the tool and version, data boundary, fallback condition, verification burden and any sign of native-skill interference. A combined result is valuable only when it beats the better safe baseline on the declared outcome without hiding new costs or loss of agency.