https://github.com/ai-domain-data/spec/

The idea: use infrastructure you already control (DNS + HTTPS) to publish one small JSON document and let assistants consume it directly.

Why: AI models routinely misinterpret websites because “who we are” is scattered across schema.org fragments, social profiles, JSON-LD, and other unstructured mediums.

AIDD gives every domain a single, self-hosted source of truth.

You publish the same JSON payload in two places:

https://<domain>/.well-known/domain-profile.json

_ai.<domain> TXT = ai-json=<base64(JSON)> (standard Base64, ≤255-character segments)

v0.1 schema (intentionally minimal):

{ "spec": "https://ai-domain-data.org/spec/v0.1", "name": "Example Publisher", "description": "Your official blurb in one or two sentences.", "website": "https://example.com", "logo": "https://example.com/logo.svg", "contact": "[email protected]", "entity_type": "business | project | personal | nonprofit | ..." }

Tooling (all open source):

Record generator & DNS/Base64 helper: https://www.ai-domain-data.org/

AI visibility checker: validates both the HTTPS and DNS sources

GitHub Action: uses: ai-domain-data/ai-domain-data-validate-action@v0 (runs aidd validate on every PR so your record doesn’t drift)

Resolver SDK (Node/TypeScript): HTTPS-first, DNS fallback, strict schema validation

WordPress plugin & Cloudflare Worker: built; publishing to their marketplaces underway

Everything lives in one repo:

Repo: https://github.com/ai-domain-data/spec

Feedback I’d appreciate:

Does the v0.1 surface area feel right? (What should wait for v0.2?)

Does the DNS/HTTPS dual-publication pattern make sense?

Are the docs/tooling clear enough for real-world adoption?

What would make you or your organization more likely to publish a self-hosted identity profile?

Thanks for taking a look!