Define agent-friendly public discovery surfaces

[arieldiaz]

Summary

• define a reusable public discovery contract for Humanware OS domain surfaces
• require canonical metadata, semantic extraction, appropriate structured data, and generated Markdown alternates where one durable source exists
• define llms.txt, robots.txt, and sitemap responsibilities
• keep human-facing revision and utility routes navigable while excluding them from the index surface with noindex,follow, canonical targets, and sitemap omission
• add release verification for discovery metadata, utility-route policy, and extracted primary text

Verification

• docs/domain-surface.md remains within its 1,000-word budget at 853 words
• git diff --check passes

Context

This extracts the reusable contract proven in arieldiaz/arieldiazcom#101 without importing site-specific content or implementation details into the framework.