This document clarifies what the CPCR Four Pillars repository does and does not publish.
This repository does publish:
- normative text for the CPCR governor (
CPCR_Spec_v1.0.md) - minimal branch-rights disclosure text (
NITT_Branch_Rights_Disclosure_v1.0.md) - conceptual framing for computability and refusal states
- a state-based Ethical Risk Index
- minimal JSON Schemas for:
- CPCR conflict records
- IRST recursive audits
- CTGS trap patterns
- PLANT-COMMONS audits
- ethical risk records
- example JSON records that satisfy those schemas
- hash and notarization conventions for document integrity
These are interfaces and governance structures, not implementation code.
This repository does not publish:
- calibration run bundles or tuning data
- runtime telemetry from any enforcement engine
- pipeline orchestration logic or internal control flow
- SparkVault hierarchy details
- TraceStack / parsing-engine internals
- Spark Lattice or SparkO$ internal mechanics
Those remain private to Spark鈥檚 own tooling and governance experiments.
Implementations may be:
- fully manual (human auditors following checklists)
- semi-automated (tools that help log and validate records)
- engine-driven (internal pipelines)
CPCR does not dictate which you must use. It requires only that:
- conflict records conform to the schemas, and
- pillar priorities and refusal semantics are respected.