---
type: "term"
title: "Semantic Guardrails (SG-1 through SG-7)"
description: "Normative rules that prevent V2.1 operators (transit boundary, intra system boundary, unresolved step) from drifting the classification model."
resource: "tlctc:term:semantic-guardrails-sg-1-through-sg-7"
tags:
  - "glossary"
---
# Semantic Guardrails (SG-1 through SG-7)

Normative rules that prevent V2.1 operators (transit boundary, intra-system boundary, unresolved-step) from drifting the classification model. They enforce that: classification is cause-first (SG-1), topology is not classification (SG-2), annotations are subordinate to clusters (SG-3), effects are not threats (SG-4), actors are not threats (SG-5), distinct exploits require distinct steps (SG-6), and stripping V2.1 annotations must leave a valid cluster sequence (SG-7).

**Reference:** §4.2.4 (Semantic Guardrails)