G-CIP stands for Contextual Integrity Protocol.
It is the core governance mechanism of the G-Master Factory (introduced in its v4.0.0-GOLD version), designed to deterministically govern AI agents and completely eliminate “hallucinations” (Context Drift or Bleeding).
This protocol structures and forces the AI to validate the context across 3 layers (levels of depth) before executing any action or state transition within the orchestrator:
- Layer L0 (Strategic): The methodological root (files like the Factory’s
state.json, the manual, and the constitution). It guarantees that we are in the correct phase of the lifecycle. - Layer L1 (Module/Directory): Rules, local READMEs, and specific workflows for the active subproject.
- Layer L2 (File): Precision at the code or document level (the Dossier, the Execution Guide, and Atomic Code Diffs).
In short, the G-CIP is the shield ensuring that no Artificial Intelligence proposes or modifies code (or text) without relying on pre-approved, documented evidence at higher architectural levels. It is what shifts the methodology from “probabilistic” to “deterministic”.
