Field data for Open Questions #5 & #6 (commerce mapping, robots defaults) — from 27 production storefronts

[vlloic]

Hi — I operate AgentReady at agentready.market, a deterministic agent-readiness audit for e-commerce stores (same name as this spec by coincidence; unaffiliated with it and with the other "AgentReady"-named products — we'd rather implement the standard that shares our name than compete with its vocabulary). We track the v0.0.1 draft surface by surface with its stable IDs: https://www.agentready.market/conformance

Because we scan production storefronts daily, we can put field numbers on two of the draft's Open Questions:

#5 — commerce protocol mapping (AR-COMM-*). In our July 2026 batch of 27 DTC storefronts (one vertical, public reports), adoption of x402 / UCP / ACP on real stores was effectively zero — while baseline machine-readability (Product JSON-LD, structured price/stock) splits the market sharply (median 56/100, bimodal). Suggestion: at v1.0.0, make the payment protocols SHOULD/MAY on the storefront surface (or gate them on "surfaces that already accept machine-initiated payment", as AR-COMM-01 hints), and keep the MUSTs at the layer stores actually fail today: structured offers and stable, addressable product URLs (AR-CONT-01/03). A MUST on x402/UCP today would fail ~100% of production storefronts and make the storefront surface unadoptable. Study with per-store data: https://www.agentready.market/research/state-of-agentic-commerce-supplements

#6 — robots.txt defaults (AR-DISC-01). Our audit treats a missing robots.txt as a permissive default (that's how every major AI crawler behaves in practice) and "declares a policy" as the improvement step. Field note: the failure mode that actually locks agents out isn't the missing file — it's the copied-template robots.txt that blocks AI user-agents by accident; one large brand in our batch scores 4/100 for exactly that. Recommendation: permissive default; declared AI policy as SHOULD.

If it helps, we're glad to draft the storefront surface requirements (offers, variant resolution, checkout-rail discovery) in the draft's stable-ID format as a PR against spec/draft.md — this is the layer we audit all day. And if an implementations/ecosystem list ever lands in the repo, we'd love to be on it.

[vlloic]

Small follow-up rather than a bump — here are the three link **TBD** items in
§5 Commerce, filled in, so you can paste them straight into spec/draft.md
without a PR round-trip:

• ACP (Agentic Commerce Protocol) — https://www.agenticcommerce.dev/ —
  spec: https://github.com/agentic-commerce-protocol/agentic-commerce-protocol
  (maintained by OpenAI and Stripe, Apache-2.0)
• UCP (Universal Commerce Protocol) — https://ucp.dev/ —
  spec: https://github.com/universal-commerce-protocol/ucp
  (Google with ecosystem partners, Apache-2.0)
• MPP (Machine Payments Protocol) — https://mpp.dev/ —
  spec: https://github.com/tempoxyz/mpp-specs (Tempo and Stripe; proposed to the
  IETF as the Payment HTTP authentication scheme, draft-httpauth-payment-00)

One thing that may be worth a line in the same section: x402 moved under the
Linux Foundation in July 2026 (x402 Foundation), so its governance story is now
different from the other three.

Happy to send any of this as a PR instead if you prefer — I just didn't want to
make you review a fork for four bullet points.

And the standing offer from above still holds: if an implementations/ecosystem
list ever lands in the repo, we'd be glad to be the first entry and to draft the
format. Our self-audit against the v0.0.1 draft, ID by ID, is public at
https://www.agentready.market/conformance