Summary
AI-Ready Web makes UAIX.org a practical reference for websites that must stay excellent for humans while becoming readable, discoverable, actionable, secure, private, auditable, and future-adaptable for AI systems. The audit finds that UAIX already has a strong baseline: clean locale routes, semantic public pages, robots.txt, XML sitemaps, llms.txt, llms-full.txt, .well-known manifests, OpenAPI, schemas, examples, validator surfaces, no-op rules, and capability ladders. The missing piece was a single AI-Ready Web program that explains how other sites should assemble those pieces without confusing stable standards, configuration-specific protocols, proposals, and research-track ideas.
Current UAIX route inventory
Source inventory now treats this page family as the canonical UAIX AI-Ready Web entry point. The supporting route inventory is available as JSON at ai-ready-web-route-inventory.json. It records current source routes, sampled live routes, known source/live package drift, static discovery files, and the planned publication boundary.
- Human pages: AI-Ready Web index plus seven volumes.
- Machine files: requirement registry, maturity register, route inventory, manifest schema/example, readiness-result schema/example.
- Existing baselines reused: Agent Compatibility, Capability Ladder, UAI-1, Validator, Launch Readiness, Privacy and Data, and Accessibility.
Live/source finding
As of the June 21, 2026 implementation pass, sampled live UAIX.org routes return HTTP 200 for the homepage, sitemap, AI Memory Package Wizard, and discovery files. Live .well-known/uaix.json still reports package version 3.125.0 while local source discovery and UAIX-Publish/ROOT-FILES discovery artifacts are 3.132.0. Live llms.txt reports 3.132.0 but still needs stale NuGet wording cleared. This page family is source-ready after release checks, but it is not live-complete until the publish folder is regenerated from current source, the coordinated production upload/root copy is applied, caches are refreshed, and live verification passes.
Risk register
| Risk | Why it matters | Required handling |
|---|---|---|
| Autonomy washing | Agents may infer write permission, safety proof, certification, or runtime authority from broad AI-ready language. | Use explicit support boundaries, no-op rules, and evidence-backed maturity labels. |
| Accessibility-tree failure | Agents and assistive technologies both rely on programmatic names, roles, headings, forms, and stable focus order. | Make WCAG 2.2, semantic HTML, keyboard/focus behavior, and stable layout the baseline. |
| Speculation drift | Emerging protocols can be useful but should not be described as stable standards before they are. | Keep MCP/A2A implementation-specific, and keep llms.txt/WebMCP/TDMRep-style signals labeled by maturity. |
| Privacy leakage | Machine files can expose more operational detail than intended. | Publish public-safe manifests only; exclude secrets, private endpoints, PII, credentials, and unreviewed logs. |
| Agent route guessing | Agents can attempt hidden endpoints or destructive actions when routes are ambiguous. | Declare highest-supported path, lowest safe fallback, human review URL, and no-op behavior. |
Volume map
Machine-readable audit files
- Requirement registry JSONStable ARW requirement identifiers, tests, evidence, and anti-patterns.
- Maturity register JSONCurrent stable, configuration-specific, proposal, research, and unsupported mechanisms.
- Route inventory JSONSource/live audit facts, publication boundary, and route exposure plan.
- AI-Ready site manifest schemaPortable declaration for discovery, capabilities, policies, evidence, and support boundaries.
- AI-Ready site manifest exampleConcrete UAIX-flavored example without claiming hosted runtime execution.
- Readiness result schemaAssessment result model for automated checks plus manual review evidence.
- Readiness result exampleExample scoring packet with warnings, blockers, and no certification claim.