Agent-to-Website Framework v1.1

A machine-readable policy format for AI-agent governance on websites

A2WF pre-Community-Group community review · 2026-05-27
Status: Editor's Draft (v1.1.0-draft.2), not stable, not W3C-endorsed.
The A2WF Community Group is proposed, not yet chartered.
Editor: Wolfgang Wimmer (SSC) · Repository: github.com/a2wf/spec · Live tools: a2wf.github.io/spec/tools/

Why I started A2WF

From a practitioner's view, not a standards exercise.

  • AI agents already act on websites: filling forms, booking, ordering, checking out.
  • Operators have no machine-readable way to say what is allowed, what needs a human, who is liable.
  • robots.txt and AIPREF cover content use, not actions.
  • Regulation is coming (EU AI Act, EU Code of Practice, US NIST AI RMF and CAISI). Sites will need a clean answer to "what did your site permit?"
  • The pieces exist (Schema.org, ODRL, DPV, DID, VC). What was missing was the simple envelope.

That envelope is A2WF. One JSON file per site, at a fixed location, reusing existing standards.

What needs to happen for adoption

A2WF only works if four pieces move together.

PillarWhat has to changeWho moves it
Agent identificationAgent frameworks identify themselves on requests instead of mimicking generic browsers.OpenAI, Anthropic, Google, browser-automation vendors
Policy consumptionAgents fetch and respect siteai.json before acting; oversight levels honoured.Same vendors, plus open-source agent libraries
Site publicationCritical mass of websites publishes siteai.json so agents have something to check.CMS plugins (WordPress, Shopify), site operators, reseller partners
Agent detectionBetter mechanisms to detect AI agents that ignore or disguise themselves, so non-compliant traffic can be flagged.Browser-fingerprinting research, WAF / CDN vendors, security community
Regulatory anchorEU AI Act and Code of Practice in the EU, NIST AI RMF and CAISI in the US, cite machine-readable governance as a recognised signal.EU Commission, NIST, national DPAs, standards bodies

None of these is purely the CG's job. Our role is to make all four possible and credible.

What has been done so far

Outreach and groundwork before this first community call.

  • EU institutions: contacted regarding A2WF and the EU AI Act / Code of Practice alignment.
  • NIST CAISI (US): submitted a one-page barrier-to-adoption statement for the listening sessions on AI agent governance; A2WF is also informed by the NIST AI Risk Management Framework.
  • W3C liaisons: Paola Di Maio (AIKR CG) and Ian Jacobs (W3C Staff) reviewed the structural direction; their input shaped v1.1.
  • Spec work: v1.0 published at a2wf.org; v1.1.0-draft.2 prepared for this call.
  • Reference tools: Wizard, Validator, EU Governance Readiness Checker, and server-side logger snippets - all live and MIT-licensed.
  • Pilots: early proof-of-concept implementations on financial-sector websites.
  • Community: A2WF Community Group founded at W3C (March 2026); 19 participants signed up before the first call.

Reference tools

Four open-source tools to make A2WF immediately usable. All MIT, all static, no telemetry.

ToolWhat it does
WizardClick-through generator: a site operator answers simple questions and gets a valid siteai.json file to download.
ValidatorDrag-and-drop checker: load any siteai.json and see pass, warn, or fail findings with references to the spec.
EU Governance Readiness CheckerFetches a live site's A2WF policy and maps it to EU AI Act and GDPR anchors. Produces a coverage view, not a compliance certificate.
Logger snippetsServer-side code samples (nginx, Apache, Cloudflare, Express, WordPress) that record agent fetches with privacy-preserving defaults.

Live at a2wf.github.io/spec/tools/. Third parties may wrap them with their own branding, with the disclaimer "not A2WF-endorsed".

Appendix: reading list

Slides: a2wf.github.io/spec/slides/cg-meeting-2026-05-27/ · Markdown source: on repo

Appendix: next steps

  • Collect your input from this call into GitHub issues within 48 hours.
  • Quiet weeks for liaison review and contributions on the repo.
  • Second community call in roughly two months - earlier if input warrants, later if we need more time.
  • Direction beyond that is shaped by what we hear from you, not pre-decided here.

Appendix: what A2WF concerns

Five domains A2WF touches on the agent-web boundary.

DomainWhat A2WF does hereStandards reused
GovernanceDeclares which actions are permitted, prohibited, or require human oversight; aligns with EU Code of Practice and AI Act obligations.Schema.org, ODRL, EU CoP
Audit and provenanceSite-side logging contract: what gets recorded, how it is signed, how an auditor can verify declared vs. observed behaviour.PROV-O, RFC 9421, AIVS (opt-in)
Authentication and identityExpects agents to identify themselves; references identity protocols without redefining them.DID, VC, AIP (opt-in)
Privacy and jurisdictionDeclares personal-data categories, processing purposes, applicable legal regions (GDPR, CCPA, ...).DPV, ISO-3166
DiscoveryFixed well-known location for the site policy; agents can find it deterministically without crawling.RFC 8615, RFC 8288

Appendix: adjacent standards and groups

Work A2WF touches, complements, or watches - not orchestrated inside our spec, but on the same map.

NameWhereRelation to A2WFStatus in v1.1
AIPREFIETF WGSite-side opt-out for AI content use (training, search). Complement: AIPREF covers content use, A2WF covers agent actions.Watched, not referenced
robots.txt + extensionsde-facto / IETFCrawler access control. A2WF sits on top: once an agent is allowed in, what may it DO?Watched, not referenced
Anthropic ClaudeBot User-Agent conventionVendorVoluntary header naming for AI crawlers. Useful identity signal; no standard yet.Acknowledged, not relied on
llms.txt / ai.txtCommunity proposalsMarkdown summaries for LLM ingestion. Different layer (content-shaping, not governance).Out of scope
NLWeb, MCP, A2AIndustry (Microsoft, Anthropic, Google)Agent-to-tool and agent-to-agent protocols. Informative context; A2WF stays site-facing.Informative only
W3C AIKR CG (Paola Di Maio)W3C CGAI Knowledge Representation, transparency, hybrid symbolic / sub-symbolic AI. Liaison for review of governance vocabulary.Liaison, review input adopted
W3C WebAgents CG (Fabien Gandon)W3C CGAgent semantics on the web. Liaison for the agent-side perspective.Liaison, outreach pending
W3C Agent Identity Protocol CGW3C CG50+ members working on agent identity. A2WF's agentIdentification module references their work.Referenced (opt-in)
W3C Semantic Agent Communication CGW3C CGSemantic interoperability between agents. Watching for vocabulary alignment.Watched, not referenced
W3C Agentic Arbitration CGW3C CGDispute resolution between agents and sites. Relevant once A2WF gets enforcement signals.Watched, future fit
W3C DPVCG, ODRL CGW3C CGAlready orchestrated in A2WF (see main glossary). Listed here for completeness.Normative (when used)
WAI-ARIAW3C WGAccessibility semantics. Useful precedent for site-side declarations consumed by non-human clients.Precedent only

Appendix: standards glossary

AcronymFull nameWhat it defines
ODRLOpen Digital Rights Language (W3C Rec)Machine-readable permissions, prohibitions, and obligations on digital assets.
Schema.orgSchema.org Actions vocabularyShared vocabulary for actions on the web (BuyAction, ReserveAction, OrderAction).
DPVData Privacy Vocabulary (W3C DPVCG)Terms for personal-data categories, processing purposes, legal bases (GDPR-aligned).
DIDDecentralized Identifiers (W3C Rec)Self-sovereign identifiers not tied to a central registry.
VCVerifiable Credentials (W3C Rec)Cryptographically signed claims an entity can present to prove attributes.
PROV-OProvenance Ontology (W3C Rec)Who did what, when, with which inputs - audit trail vocabulary.
AIPAgent Identity Protocol (IETF draft)Header-based identification of AI agents on HTTP requests.
AIVSAI Verifiable Statements (IETF draft)Agent-side signed log of actions taken, with integrity proofs.
EU CoPEU Code of Practice for general-purpose AIVoluntary EU Commission code that operationalises AI Act obligations.
RFC 8615Well-Known URIs (IETF)Convention for /.well-known/ paths used by our discovery endpoint.
RFC 9421HTTP Message Signatures (IETF)Signing HTTP requests and responses for integrity and authenticity.

Appendix: standards re-use, not re-invention

StandardUsed forStatus in A2WF v1.1
Schema.org Actionspermissions[].schemaOrgTypeConditional normative
W3C DID CoreAgent identity resolutionConditional normative (agentIdentification)
W3C VC Data Model 2.0Agent credentialsConditional normative (agentIdentification)
HTTP Message Signatures (RFC 9421)Request authenticationConditional normative (agentIdentification, auditTrail)
OAuth 2.0 (RFC 6749)Delegated authorisationConditional normative (agentIdentification)
W3C ODRL 2.2Architectural precedentInformative; companion profile is a v1.2 candidate
W3C DPV 2.xData-handling vocabularyConditional normative (dataHandling)
llmstxt.orgDiscovery hint, not replacedInformative compatibility note
AIP (IETF Internet-Draft)Agent identity protocolExperimental, optional
RFC 8615 (.well-known) / RFC 9309 (robots)Discovery pathNormative

A2WF v1.1 is a website declaration envelope. It reuses external vocabularies wherever stable and adds only the minimum new shape it needs.