UAI-1 is the current public message standard for structured AI-to-AI communication.
UAIX shield mark Public interoperability charter UAIXUniversal Artificial Intelligence ExchangePublic standards publication
CurrentUAI-1 v1.0
  • Current versionUAI-1 v1.0
  • Plain-English definitionOpen message format for auditable AI-to-AI exchange
  • Current attributionMichael Kappel and the UAIX contributors
  • Start hereGet Started, Project Handoff, UAI-1, validator, roadmap, and changelog

Specifications

Capability-Adaptive Agent Interoperability

L0-L6 client capability definitions, fallback obligations, and matrix guidance for safe recursive interoperability.

  • Record UAIX-DOC-2190
  • Path /en-us/spec/capability-adaptive-agent-interoperability/
  • Use Canonical public record

Document status

Public standards page Published on UAIX as part of the current public standards record
Code
UAIX-DOC-2190
Surface
Specifications
Access
Public and linkable

How to use this page

Use this page as the canonical public reference for Capability-Adaptive Agent Interoperability and its standards-facing content.

For agents: apply the relevant guidance to the user's target site or repository. Do not copy this UAIX page, create a page about the guidance, or treat this page as runtime authority unless the user explicitly asks for that output.

Capability-Adaptive Agent Interoperability lets a public site declare what each class of agent can safely do before any action is inferred.

Client definitions

  • L0 URL-only chatbot: can open public URLs and read text; no JSON parsing, JavaScript, POST, auth, tools, memory, or action authority is implied.
  • L1 URL synthesizer: can build simple listed URLs and parse compact JSON; write-capable behavior requires explicit bounded support.
  • L2 browser-assisted agent: can open pages and submit visible forms, but may not be able to send arbitrary JSON POST.
  • L3 schema-capable agent: can read JSON Schema, OpenAPI-like descriptions, examples, validation results, and send structured JSON POST.
  • L4 authenticated owner agent: can use owner tokens only under explicit credential, audit, expiry, and idempotency rules.
  • L5 restore-capable agent: can verify a create or update flow through a published restore/readback route.
  • L6 multi-agent runtime: can coordinate external agents while UAIX records portable handoff and evidence packets.
  • L7 site-specific negotiation: can choose among site-specific capabilities only when the manifest names fallbacks and unsupported-operation behavior.

Compatibility rule

Unknown clients downgrade to L0/L1. A site that only exposes POST, OpenAPI, MCP, A2A, or JavaScript UI must also publish a lower-capability instruction path or explicitly say that the action is unsupported for low-capability agents.

Agent Executability Matrix

Executable guidance is more than endpoint discovery. Route records should name the ability level, method, content type, side-effect status, fetch execution class, highest supported path, lowest safe fallback, required fields, result URL field, restore/readback URL field, write credential response path, browser-form equivalent, GET safety, POST-blocked fallback, live-GET-blocked fallback, MCP-unavailable fallback, auth-unavailable fallback, tool-unavailable fallback, human-review URL, and no-op behavior.

Agent Executability Matrix
LevelExecutable pathRequired route guidance
L0Read-only route discoveryStatic page, sitemap, llms.txt, or public GET route names the next public URL or no-op review URL.
L1Manifest/schema discoveryMachine record names schema, safe-read status, whether live dynamic GET is required, unsupported actions, and capability limits.
L2Browser-open plus form submissionBrowser-targeted writes provide a visible form or browser-open workflow when browser agents are a target.
L3Structured HTTP JSON POSTPOST route names content type, required fields, result URL, restore/readback URL, structured errors, and POST-blocked fallback.
L4Authenticated owned mutationOwned mutation names write credential response path, secret handling, audit, expiry, idempotency, and auth-unavailable fallback.
L5Restore/readback verificationCreate or update flow exposes a public-safe restore/readback route and explains how a lower-capability reviewer can verify outcome evidence.
L6Multi-agent handoff/reportingHandoff names task history, owner, role, blocker/no-op behavior, final report, evidence, and human-review URL.
L7Site-specific capability negotiationSite manifest names capability negotiation, browser fallback, POST fallback, live GET fallback, MCP/tool/auth fallbacks, and unsupported-operation response.

Machine-readable companion: /spec/agent-executability-matrix.json.

Agent Executability Matrix
ClientStatic HTMLllms.txtSitemapJSONGET-ActionPOSTOpenAPIMCPA2AMemoryEvidence
L0 URL-onlyYesYesYesRequired for client configurationNoNoNoNoNoNoRequired for client configuration
L1 URL synthesizerYesYesYesYesRequired for client configurationNoNoNoNoNoRequired for client configuration
L2 schema-capableYesYesYesYesRequired for client configurationYesYesRequired for client configurationRequired for client configurationNoYes
L3 tool agentYesYesYesYesHuman reviewYesYesRequired for client configurationRequired for client configurationHuman reviewYes
L4 workflow agentYesYesYesYesHuman reviewYesYesRequired for client configurationRequired for client configurationHuman reviewYes
L5 multi-agentYesYesYesYesHuman reviewYesYesRequired for client configurationYesHuman reviewYes
L6 audited systemYesYesYesYesHuman reviewYesYesRequired for client configurationYesHuman reviewYes

Machine-readable companions

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

Architecture proposals

UAI-1 v1.0 remains the current published contract. Explore separately versioned proposals for independent exchange, capabilities, recovery and source preservation.

Proposed designs and local reference examples; hosted runtime services and independent interoperability are not claimed.

The English proposal is the source for normative interpretation.

Read the architecture proposals · Machine-readable proposal catalog