Summary
AI-Ready Web is the UAIX program for websites that need to remain excellent for humans while becoming reliable for AI systems. It defines how to make pages accessible, content discoverable, APIs deterministic, capability claims bounded, delegation auditable, privacy preserved, and future standards adoption honest.
Scope
The program complements UAI-1. UAI-1 is the portable public exchange, evidence, memory, trust-declaration, and handoff layer. HTTP APIs and OpenAPI describe route-level interfaces. MCP describes model/tool/resource integration in compatible host environments. A2A or other protocols may handle agent discovery, delegation, and task coordination when a site actually implements them. AI-Ready Web records how those layers fit together without merging them.
Operating principles
- Human-first, agent-compatible: do not sacrifice people, accessibility, or clarity for bots.
- Stable standards before speculative protocols: label every mechanism by maturity.
- Evidence over assertion: current support requires page copy, machine artifact, test, and release evidence.
- Least privilege and no-op safety: agents should use the lowest safe route and stop when unsupported.
- Vendor neutrality: the program should work across browsers, frameworks, runtimes, and agent vendors.
- One source of truth: human pages, manifests, schemas, examples, validators, and release notes must agree.
Seven-volume program
Current maturity model
| Status | Use it for | Examples | UAIX rule |
|---|---|---|---|
| Stable baseline | Requirements that should be implemented before any agent-specific claim. | WCAG 2.2, semantic HTML, HTTP semantics, robots.txt, sitemap, JSON Schema, OpenAPI, Problem Details, Trace Context, JSON-LD. | May be required language when relevant and tested. |
| Current optional | Useful capabilities with real implementations but environment-specific adoption. | MCP in compatible hosts, A2A where supported, signed non-human principal flows, structured alternate representations. | Label as supported only when the local implementation has public evidence. |
| Proposal or community convention | Helpful discovery or policy signals that are not formal web standards. | llms.txt, markdown mirrors, agent preference files, TDMRep-style rights signals. | Use as advisory signals and never as the only source of authority. |
| Research track | Ideas to monitor without current support claims. | WebMCP/browser-native tool declarations, DNS-based agent discovery, autonomous agent commerce credentials beyond published APIs. | Keep in roadmap language until specifications, implementations, tests, and release evidence exist. |
| Unsupported | Claims UAIX must not imply. | Hosted runtime execution, automatic repository writes, hidden credential validation, certification, endorsement, safety proof, consciousness proof. | Block or rewrite the claim. |
Machine-readable files
- Requirement registry JSONStable ARW requirement identifiers, tests, evidence, and anti-patterns.
- Maturity register JSONCurrent stable, current optional, 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.
Support boundary
AI-Ready Web records are public guidance, schemas, examples, and review evidence. They do not authorize scraping, bypass local policy, validate credentials, grant tools, execute workflows, certify vendors, endorse agents, prove safety, or prove consciousness. Unsupported actions should return no-op plus human review.