Changelog

All notable changes to the AI Visibility Definition are documented here. The current version is 1.9.0.

This project follows Semantic Versioning and the Keep a Changelog format. Subscribe to the Atom feed for version updates.


1.9.0 — 11 May 2026

Phase 6 of the spec standardisation work. Adds the long-term / forward-looking layer: a public roadmap, formal rules for extensions, internationalisation and accessibility guidance, the Discovery: directive in robots-ai.txt, a formal media-type stance, and an expanded integrity / signing roadmap. The Discovery: directive is the only normative addition to publisher behaviour; all other additions are forward-looking documentation.

Added

  • Roadmap. Theme-pegged forward plan with status flags (Active, Next, Future, On hold). Intentionally not version-pegged.
  • Extensions. Normative rules for experimental x--prefixed files, extension fields inside core JSON, and the six-step promotion path from experimental to core.
  • Internationalisation & Accessibility. Multi-language publication, locale-tagged identity fields, RTL languages, non-Latin scripts, and accessibility of llms.html.
  • The Discovery: directive in the robots-ai.txt specification. Publishers MAY advertise AI Discovery Files present on the same host. Solves the cold-start problem for consumers that do not probe root-level paths blindly.
  • Formal media-type stance at HTTP behaviour section 10. Reuse existing text/plain, text/markdown, text/html, and application/json rather than introducing bespoke types.
  • Expanded file integrity and signing section at security and privacy section 6. Four candidate signing mechanisms, six cross-cutting concerns, and interim guidance for publishers and consumers while signing remains On hold for a future MAJOR release.

Changed

  • All 10 AI Discovery File specifications bumped MINOR. Eight specs 1.6.0 to 1.7.0; two (faq-ai-txt, developer-ai-txt) 1.7.0 to 1.8.0.
  • All 3 guides bumped to 1.6.0.
  • Master definition bumped to 1.9.0.
  • Registry data file bumped to 1.5.0.

Phase reference

This release completes Phase 6 of the spec standardisation work documented in docs/plans/2026-05-10-spec-standardisation-implementation.md. Phases 1 through 6 are now complete; the original ChatGPT review brief is fulfilled.

1.8.0 — 11 May 2026

Phase 5 of the spec standardisation work. Adds the positioning layer: where the AI Discovery Files specification sits in the wider standards landscape, who has implemented it, how it stays backwards-compatible with the llmstxt.org convention, and how the host-scoping rule applies across domains and subdomains.

Added

  • Relationship to Other Standards. Positions the specification against ten neighbours: llmstxt.org, IETF AI Preferences, robots.txt and RFC 9309, Schema.org, Common Crawl and AI crawler user agents, W3C AI Use Disclosure, BCP 14, JSON Schema 2020-12, Semantic Versioning 2.0.0, and other web-discovery conventions.
  • Implementations. Public record of conformant implementations, modelled on the IETF implementation-report pattern. Lists the four implementation roles (Publisher, Validator, Registry, Consumer) and the five current implementations (WordPress plugin, AI Visibility Checker, AI Visible Directory, manual templates, Service Pack).
  • llmstxt.org backward-compatibility statement in the llms.txt specification. An llmstxt.org-conformant llms.txt file is a valid llms.txt at the Essential conformance class under this specification. Publishers who follow the llmstxt.org convention need no migration to claim Essential conformance.
  • Multi-domain and subdomain scoping rule in the llms.txt and identity.json specifications. Files are host-scoped per RFC 3986; cross-host identity MUST be asserted via mutual sameAs declarations.
  • Standards Framework section on the specifications hub and the spec sidebar. Surfaces the Phase 2-5 framework pages (conformance, conventions, registry, versioning, governance, security and privacy, HTTP behaviour, processing model, consumer guidance, test vectors, related standards, implementations) as a dedicated block.

Changed

  • All 10 AI Discovery File specifications bumped MINOR. Eight specs 1.5.0 to 1.6.0; two (faq-ai-txt, developer-ai-txt) 1.6.0 to 1.7.0.
  • All 3 guides bumped to 1.5.0.
  • Master definition bumped to 1.8.0.
  • Registry data file bumped to 1.4.0.

Phase reference

This release completes Phase 5 of the spec standardisation work documented in docs/plans/2026-05-10-spec-standardisation-implementation.md.

1.7.0 — 11 May 2026

Phase 4 of the spec standardisation work. Adds the implementer-facing depth layer: a formal processing model for conformant consumers, guidance on what AI systems should do with publisher signals, a canonical test-vector framing, and reference-implementation framing on the AI Visibility Checker. No normative requirements changed.

Added

  • Processing Model. Algorithmic specification covering the seven stages a conformant consumer applies: discover, fetch, validate, resolve identity, resolve permissions, detect contradictions, emit. Given the same input, two conformant implementations must produce equivalent output.
  • AI Consumer Guidance. What AI systems should do with AI Discovery Files. Covers training-time use, retrieval-time use, citation conventions, conversational behaviour, contradiction handling, and defensive posture.
  • Test Vectors framing. Documents where the canonical test suite lives (the ai-discovery-files spec repository), how each test case is structured (input files, expected.json, readme), the requirement-ID convention, and how validator authors apply the suite.
  • Reference-implementation framing on the AI Visibility Checker. The checker hero now formally states it is the reference implementation of the AI Discovery Files specification.

Changed

  • All 10 AI Discovery File specifications bumped MINOR. Eight specs 1.4.0 to 1.5.0; two (faq-ai-txt, developer-ai-txt) 1.5.0 to 1.6.0.
  • All 3 guides bumped to 1.4.0.
  • Master definition bumped to 1.7.0.
  • Registry data file bumped to 1.3.0.

Phase reference

This release completes Phase 4 of the spec standardisation work documented in docs/plans/2026-05-10-spec-standardisation-implementation.md.

1.6.0 — 11 May 2026

Phase 3 of the spec standardisation work. Adds the process, policy, and security layer that surrounds every specification: versioning and deprecation policy, governance and editorial process, security and privacy considerations, and HTTP behaviour. No normative requirements changed; every addition is documentation surfacing rules and contracts that the rest of the specification family already assumed.

Added

  • Versioning and Deprecation Policy. Single source for Semantic Versioning 2.0.0 commitments (MAJOR, MINOR, PATCH), the deprecation timeline (minimum 12-month window between Deprecated and Retired), and the specification lifecycle from Draft through Stable to Deprecated to Retired.
  • Governance and Editorial Process. Documents the maintainer, the five-status proposal lifecycle (Proposed, Accepted, Published, Stable, Deprecated), the editorial process from capture to publication, working principles, and the contribution path through the project GitHub repository.
  • Security and Privacy Considerations. Cross-cutting guidance: trust model (HTTPS host as anchor, no signing yet), content-injection patterns publishers should avoid, GDPR considerations for personal data in publisher files, the access-control boundary (AI Discovery Files are advisory, not enforcement), logging and fingerprinting implications, and the forthcoming file-integrity primitives.
  • HTTP Behaviour. The HTTP-protocol-level contract every AI Discovery File specification assumes: HTTPS expected, status code expectations, redirect handling with a 5-hop maximum, soft-404 detection, query-string treatment, content negotiation expectations, cache headers, rate-limit guidance, and CORS.

Changed

  • All 10 AI Discovery File specifications bumped MINOR. Eight specs (llms-txt, llm-txt, llms-html, ai-txt, ai-json, identity-json, brand-txt, robots-ai-txt) 1.3.0 to 1.4.0; two (faq-ai-txt, developer-ai-txt) 1.4.0 to 1.5.0. New versionHistory[] entry on every spec PHP page, JSON twin, and YAML twin.
  • All 3 guides (quick-start, interoperability, validation) bumped MINOR to 1.3.0 with matching versionHistory[] entries.
  • Master definition bumped to 1.6.0 as the umbrella release version.
  • Registry data file bumped to 1.2.0 to reflect Phase 3 spec versions in files[].version.
  • Specification Conventions page: the placeholder "forthcoming versioning policy page" note replaced with a link to the now-published versioning page. References list updated with links to all four new Phase 3 pages.

Spec affected

  • ADF-001 through ADF-010 (all bumped)
  • AV-001 through AV-006 wording unchanged

Phase reference

This release completes Phase 3 of the spec standardisation work documented in docs/plans/2026-05-10-spec-standardisation-implementation.md.

1.5.0 — 11 May 2026

Phase 2 of the spec standardisation work. Adds the conformance regime and the machine-readable infrastructure the rest of the standardisation depends on. All items shipped to the live site across batches between 2026-05-10 and 2026-05-11 as additive changes; this release is the coordinated bump that updates every spec and guide data file in lockstep.

Added

  • Conformance specification. Formally defines the three conformance classes (Essential, Recommended, Complete), publisher conformance criteria, validator conformance criteria, and the distinction between self-declared and Directory-verified conformance. Every ADF spec page now carries a Conformance section listing the classes that include the file as anchored badges.
  • Specification Registry at /specifications/registry.json. Machine-readable single source of truth listing all 10 AI Discovery Files with paths, media types, paired files, redirect targets, precedence rules, schema URLs, conformance-class membership, and current versions. Validators, plugins, and downstream tooling SHOULD load this at runtime rather than hard-coding the file list.
  • Spec meta-schema at /specifications/spec-meta.schema.json. JSON Schema 2020-12 describing the shape of every *-specification.json data file.
  • Validator-output schema at /specifications/validator-output.schema.json. Standard output format any conformant validator emits, so different validators can produce equivalent results for the same input. The AI Visibility Checker is the reference implementation.
  • Versioned JSON Schema URLs at /specifications/ai-json/v1/ai-json.schema.json and /specifications/identity-json/v1/identity-json.schema.json. Permanent URLs alongside the existing unversioned "latest" aliases. Publishers can now pin their $schema field to a URL that will never change shape. See Schema URL versioning.
  • Optional BCP 47 language declaration. Every AI Discovery File MAY declare its natural language using a BCP 47 tag. Lang: header for text files; language property for JSON files; existing <html lang> attribute for llms.html. See Language declaration in publisher files.
  • Audit script extensions. Seven new Phase 2 checks covering registry shape, registry-versus-versions.php sync, registry paths on disk, spec-meta validation of every *-specification.json, validator-output schema compilation, versioned-schema $id consistency, and BCP 47 language tag validity.

Changed

  • All 10 AI Discovery File specifications bumped MINOR. Eight specs (llms-txt, llm-txt, llms-html, ai-txt, ai-json, identity-json, brand-txt, robots-ai-txt) 1.2.0 to 1.3.0; two (faq-ai-txt, developer-ai-txt) 1.3.0 to 1.4.0. New versionHistory[] entry on every spec PHP page, JSON twin, and YAML twin.
  • All 3 guides (quick-start, interoperability, validation) bumped MINOR to 1.2.0 with matching versionHistory[] entries.
  • Master definition bumped to 1.5.0 as the umbrella release version.
  • Registry data file bumped to 1.1.0 to reflect Phase 2 spec versions in files[].version.
  • Canonical examples at root: /ai.json and /identity.json now reference the versioned schema URL (/v1/) and demonstrate the optional language field. /llms.txt demonstrates the Lang: en-GB header.

Fixed

  • /checker/templates/identity.json: corrected a pre-existing bug where the template still referenced the old 365i.co.uk/ai-visibility-definition/... URL from before the project moved to its own domain. Now references the v1 URL on ai-visibility.org.uk.

Spec affected

  • ADF-001 through ADF-010 (all bumped)
  • AV-001 through AV-006 wording unchanged

Phase reference

This release completes Phase 2 of the spec standardisation work documented in docs/plans/2026-05-10-spec-standardisation-implementation.md.

1.4.0 — 10 May 2026

Phase 1 of the spec standardisation work — moves the AI Discovery Files specification from "useful published convention" to "credible community specification" by adopting the formal positioning, RFC 2119 keyword discipline, document statuses, conventions/licensing reference pages, and a coordinated release cadence. No normative requirements changed in this release; all changes are additive or stylistic normalisation.

Added

  • Positioning statement — Published verbatim on the directory homepage, the specifications hub, and the definition page. Frames AI Discovery Files as an open, versioned community specification working towards recognised implementation-standard status.
  • /licensing/ page — Single source for the project's licensing and trademark stance: CC BY 4.0 for specification text and examples, MIT for JSON Schemas, "AI Discovery Files" as descriptive (non-trademarked) phrase free for accurate reference, AI Visibility Directory verification badges as restricted assets.
  • /specifications/conventions/ page — Single source for editorial and structural conventions used across every AI Discovery File specification. Cites RFC 2119 and RFC 8174 (BCP 14) for requirement keywords; documents the four document-status values (Draft, Stable, Deprecated, Retired); summarises Semantic Versioning 2.0.0 across PHP/JSON/YAML.
  • "Status of This Document" block — Inserted on every spec page (10), every guide page (3), and the new conventions and licensing pages. All currently Stable. Each block links to /specifications/conventions/#status-values for the canonical status definitions.
  • References section on every spec and guide page — Each spec now ends with a References section linking to the conventions page and the licensing page.
  • /licensing/ link in the site footer — Between Terms and Support.
  • Audit script (scripts/audit-spec-consistency.js) — Mechanical consistency checks across PHP/JSON/YAML data files: parse, version sync, dateModified vs versionDate, $schema discipline both ways, AV-xxx code consistency, JSON/YAML structural equivalence. Phase-gated; extends in each subsequent phase.

Changed

  • All 10 AI Discovery File specifications — Bumped MINOR (llms-txt, llm-txt, llms-html, ai-txt, ai-json, identity-json, brand-txt, robots-ai-txt: 1.1.1 → 1.2.0; faq-ai-txt, developer-ai-txt: 1.2.0 → 1.3.0). Added Status block, applied RFC 2119 keyword normalisation in normative content, added References section.
  • All 3 guides (quick-start, interoperability, validation) — Bumped MINOR to 1.1.0 with the same set of additions.
  • Master definition — Bumped to 1.4.0 as the umbrella release version. Same additions applied to the definition page.
  • RFC 2119 keyword normalisation — Across all 13 specs/guides plus their JSON/YAML twins, lowercase keywords inside normative content were normalised to uppercase per RFC 2119 / RFC 8174. Approximately 110 keyword occurrences uppercased across the codebase.

Fixed

  • faq-ai-txt-specification.yaml — Three list items with unquoted colons that YAML interpreted as mapping syntax caused parse failure or silent semantic divergence with the JSON twin. Quoted; YAML now parses correctly and matches JSON shape.
  • llm-txt-specification.yaml — Missing iis example key under implementation.methods[0].examples that exists in the JSON twin. Added.

Spec affected

  • ADF-001 through ADF-010 (all)
  • AV-001 through AV-006 wording unchanged

1.3.3 — 13 February 2026

Added

  • AI Visibility Directory CTA — Registration guidance added to all specification pages, guide pages, and content pages promoting the AI Visibility Directory with two-tier listing comparison (Basic and Full)
  • Version history entries (v1.1.1) added to all 10 AI Discovery File specification pages
  • Version history entries added to Quick Start (v1.0.2), Interoperability (v1.0.1), and Validation (v1.0.1) guides
  • Changelog link added to Specifications navigation dropdown

Changed

  • "Last Modified" date on definition page is now dynamic (uses file modification timestamp instead of hardcoded date)
  • Specification versions bumped from 1.1.0 to 1.1.1

1.3.2 — 16 January 2026

Added

  • Implementation Guides — Three new guide pages to support specification adoption:
    • Quick Start Guide — Tiered implementation path with Essential (2 files), Recommended (6 files), and Complete (10 files) tiers
    • Interoperability Guide — Conflict resolution rules and precedence hierarchy for when AI Discovery Files contain contradictory information
    • Validation Framework — Standardised test prompts and scoring methodology to verify AI systems correctly interpret AI Discovery Files
  • Guides section in specification sidebar navigation
  • Implementation Guides section on specifications hub page with callout for newcomers
  • Precedence rules and cross-references added to all 10 specification pages linking to Interoperability Guide

1.3.1 — 12 January 2026

Changed

  • Site Restructure — Converted from static HTML to PHP-based architecture with external CSS, shared includes, dynamic schema generation, specification sidebar, and client-side search
  • Dynamic dateModified using file modification timestamps
  • ISO 8601 datetime format with timezone for Google Rich Results compliance

Fixed

  • DateTime format validation errors in Google Rich Results Test

1.3.0 — 12 January 2026

Added

  • AI Discovery File Specifications — 10 individual specification pages for each core AI Discovery File format: llms.txt, llm.txt, llms.html, ai.txt, ai.json, identity.json, brand.txt, faq-ai.txt, developer-ai.txt, robots-ai.txt
  • Machine-readable specification files (JSON and YAML) for each format
  • JSON Schema files for ai.json and identity.json validation
  • Canonical example files using fictional business "Horizon Strategic Consulting"

1.2.0 — 12 January 2026

Added

  • ai.json added to core AI Discovery Files list — machine-parseable AI interaction guidance in JSON format

1.1.0 — 12 January 2026

Added

  • Frequently Asked Questions — Six questions addressing common clarifications about AI Visibility
  • FAQPage JSON-LD structured data
  • FAQ entries in machine-readable JSON and YAML formats

1.0.0 — 10 January 2026

Added

  • Initial publication of canonical definitions
  • AV-001: AI Visibility — the degree to which a website can be discovered, interpreted, and cited by AI systems
  • AV-002: AI Visibility Checking — technical verification of AI-facing infrastructure
  • AV-003: AI Discovery Files — machine-readable files for AI communication
  • AV-004: AI Visibility Tracking — ongoing measurement of AI-generated mentions
  • AV-005: AI Visibility Monitoring — real-time observation of AI behaviour
  • AV-006: AI Retrieval Testing — prompt-based observation of AI responses
  • Taxonomy establishing hierarchical relationships between terms
  • Reference implementation link to 365i AI Website Visibility Checker
  • HTML, JSON, and YAML format availability
  • Schema.org DefinedTerm structured data
Free WordPress Plugin

Generate AI Discovery Files from your dashboard

Using WordPress? Install the plugin and create all 10 files in minutes — no coding, no configuration files to edit manually.

Get the Plugin

Register in the AI Visibility Directory

Once your AI Discovery Files are published, register your website in the AI Visibility Directory — the verified registry of websites implementing AI Discovery Files. Registration validates your implementation and lists your site for AI systems and industry peers to discover.

Basic Listing

Card entry in the directory with automated file validation. Open to any site with a valid llms.txt file. No cost.

Full Listing Recommended

Dedicated profile page on the directory with dofollow backlinks to your website — a genuine SEO authority signal from a topically relevant, verified source. Includes an attribution badge and enhanced visibility.