GOV-002 · Generative Search, Scale, and Governance

Do not depend on llms.txt for Google Search

• Baseline
• IF
• THEN
• ELSE

The decision

• IF
  • a proposal claims a special AI text file is required for Google visibility
• THEN
  • reject the requirement and prioritize public crawlable indexable content
• ELSE
  • maintain such a file only for a separate system that explicitly uses it

Primary metric

eligible indexed coverage

Guardrail

maintenance distraction

Reason

Makes the decision explicit, measurable, and reversible where possible.

LAKA levels

• Baseline

Operators

• IF
• THEN
• ELSE

Governs gates

• maintainability
• compliance

Source

04-IF-ELSE-AND-OR-RULEBOOK.md — GOV-002 — Do not depend on llms.txt for Google Search

The other eleven rules in Generative Search, Scale, and Governance

• GOV-001 Keep foundational SEO as the base IF the goal includes visibility in generative search experiences Baseline
• GOV-003 Do not chunk content mechanically IF content is being split only because an AI system is assumed to need tiny sections Major
• GOV-004 Use query fan-out as a research map IF a broad request implies related information needs Major
• GOV-005 Block scaled content abuse IF content volume is generated primarily to manipulate rankings or AI responses without distinct user value Structural
• GOV-006 Measure generative visibility directly IF Search Console or another first-party surface provides generative-search reporting Baseline
• GOV-007 Do not invent AI-specific schema IF a proposal adds unsupported structured data solely for generative visibility Baseline
• GOV-008 Prepare for agentic use through usability IF browser agents or assistive systems need to interpret and operate the site Structural
• GOV-009 Expose freshness and provenance IF data may change or be reused in generated answers Major
• GOV-010 Constrain programmatic expansion IF candidate combinations exceed quality or maintenance capacity Structural
• GOV-011 Protect privacy, consent, and sensitive data IF volumetric generation or personalization uses personal, confidential, licensed, or restricted data Structural
• GOV-012 Pursue paradigm assets selectively IF a new tool, dataset, benchmark, diagnostic, or agent can complete the task far better than existing result classes Paradigm