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

Agent Compatibility

Validator Rules

Agent compatibility validator failures and warnings for GET-Action fallback, idempotency, query secrets, JavaScript-only discovery, llms.txt, POST equivalents, and no-op behavior.

  • Record UAIX-DOC-2754
  • Path /en-us/spec/validator-rules/
  • Use Canonical public record

Document status

Public standards page Published on UAIX as part of the current public standards record
Code
UAIX-DOC-2754
Surface
Agent Compatibility
Access
Public and linkable

How to use this page

Use this page as part of the current Agent Compatibility public record, then follow its linked standards pages for the next step.

Agent compatibility validation checks the lowest advertised capability first. Rich interfaces pass only when lower-capability discovery is honest and bounded.

Required failures

  • get_action_fallback_required: L0/L1 compatibility fails when a core writable action is POST-only and no documented GET fallback or no-op URL exists.
  • get_action_idempotency_required: GET-Action fails when a write-capable example lacks idempotency_key or nonce.
  • get_action_query_secret: GET-Action fails when examples put secrets, tokens, passwords, account identifiers, medical identifiers, payment data, private messages, or regulated identifiers in URLs.
  • get_action_js_only_discovery: GET-Action fails when discovery is hidden behind JavaScript-only UI.
  • llms_minimal_instruction_required: L0/L1 discovery fails when llms.txt points only to complex OpenAPI or tool specs and omits direct minimal instructions.
  • retry_loop_no_op_violation: no-op handling fails when unsupported agents are told to keep retrying or guessing.

Required warnings

  • post_equivalent_recommended: warn when a GET-Action has no POST or OpenAPI equivalent for capable clients.
  • get_action_security_order_warning: warn when examples appear before consent and security boundaries.

Test fixtures

The release publishes positive and negative fixture JSON records for these rule families under the UAI-1 examples tree and exposes the rule catalog through agent-compatibility.json.

Boundary

Agent runtimes execute. UAIX records the reviewed communication, memory, trust, evidence, and handoff boundary. UAIX.org does not execute third-party actions, certify safety, certify truth, endorse implementations, or replace local authorization policy.

Agent Compatibility Records