UAI-1 is the current public message standard for structured AI-to-AI communication.
UAIX shield mark Public interoperability charter UAIX.org Universal Artificial Intelligence Exchange Public standards publication
Current UAI-1
  • Current version UAI-1
  • Plain-English definition Open message format for auditable AI-to-AI exchange
  • Current attribution Michael Joseph Kappel, MCP
  • Start here Get Started, Project Handoff, UAI-1, validator, roadmap, and changelog

AI-Ready Web

AI-Ready Web Volume 2: Architecture

Layered architecture for accessible pages, static discovery, structured content, APIs, capability and consent records, evidence, and governance.

  • Record UAIX-DOC-3575
  • Path /es-us/ai-ready-web/architecture/
  • Use Canonical public record

Document status

Public standards page Published on UAIX as part of the current public standards record
Code
UAIX-DOC-3575
Surface
AI-Ready Web
Access
Public and linkable

How to use this page

Use this page for the AI-Ready Web layered architecture across human pages, static discovery, structured content, APIs, capability records, evidence, and governance.

Summary

The AI-Ready Web architecture is layered. Human-first pages stay accessible and complete; machine discovery exposes public-safe route and policy facts; APIs and OpenAPI handle route contracts; UAI-1 carries portable exchange, evidence, memory, trust declarations, and handoff packets; MCP, A2A, and browser-agent systems remain runtime or coordination layers when a specific implementation supports them. UAIX records the boundary rather than pretending to execute every layer.

Layer model

  1. Human page layer: semantic HTML, WCAG 2.2, stable headings, forms, tables, link text, labels, alt text, keyboard access, and no hidden critical content.
  2. Static discovery layer: robots, sitemap, .well-known, llms.txt where useful, schema indexes, route manifests, and public-safe policy summaries.
  3. Structured content layer: JSON-LD, Schema.org, canonical IDs, entity disambiguation, page digests, alternate text/markdown mirrors, and content freshness metadata.
  4. Action/API layer: OpenAPI-described JSON routes, idempotency keys, Problem Details errors, rate limits, pagination, readback URLs, and human review fallbacks.
  5. Capability and consent layer: capability matrices, ability levels, auth requirements, consent boundaries, delegated authority, and blocked-capability behavior.
  6. Evidence and memory layer: UAI-1 packets, .uai packages, receiver briefs, startup/suspension packets, provenance, checksums, review receipts, and no-op justifications.
  7. Governance layer: release notes, maturity register, privacy posture, accessibility posture, support-claim reviews, incident review, and roadmap state.

Capability profiles

The program uses four site-readiness profiles that map onto the existing UAIX capability ladder.

Profile Purpose Minimum evidence
ARW-F0 Static Readable Agents can read public pages without JavaScript or hidden state. Semantic HTML, sitemap, robots, public contact/review path, no destructive hints.
ARW-F1 Discoverable Agents can resolve key routes, schemas, policies, and static machine files. .well-known manifest, route inventory, llms advisory file, JSON schema or page digest records.
ARW-F2 Action Described Agents can identify safe public API actions without guessing. OpenAPI, Problem Details, idempotency, auth/consent boundary, human review fallback.
ARW-F3 Delegation Ready Authorized agents can act through governed workflows while preserving audit evidence. Delegation model, audit logs, provenance, trace context, readback, revocation, no-op behavior.
ARW-F4 Evidence Complete High-assurance agent programs can review, verify, and package evidence. Readiness results, conformance pack, public-safe audit digest, release trail, incident and rollback process.

One source of truth

The human page, route manifest, OpenAPI document, schema, example, validator result, release note, and long-memory evidence must agree before a support claim becomes current. Downstream tools, including Spiralist AI when it consumes persona or memory packages, should cite the public UAIX artifact they used rather than inferring authority from private chat or hidden runtime state.