ChatGPT Instant Checkout / OpenAI ACP
openai.com · Agent-Commerce Infrastructure modeIntegrated
Overall = mean of Interrupt Score 73 and Machine-Readability 90 → 82 / 100. The badge is set by the Interrupt Score alone — how scoring works.
Confidence: Verified (Documented) (0.8) — Backed by strong public-record evidence of real end-to-end behavior, without our own live probe run.
This tool doesn't involve holding funds or assets, so custody risk doesn't apply.
Machine-Readability Score
Four check groups, equal weight: machine interfaces (OpenAPI, MCP server, llms.txt), browsing efficiency, element discoverability, and discovery/agent access (retrieval-bot robots.txt access, sitemap, schema.org). Each check is tagged with how strong the real-world evidence is that it changes agent behavior — proven, plausible, or emerging — grades are honesty labels, they never weight the score. A separate signal about ease of integration — it never enters the Interrupt Score or badge above.
Findings
How scoring works →Interrupt tier: Moderate-Interrupt — derived from 1 finding; interrupt score 73 / 100. It alone drives the badge above; the headline Overall score blends it 50/50 with Machine-Readability.
Every tool starts at 100. Each interrupt found subtracts a penalty of base weight × timing × hardness × evidence — the multiplier values below come from the same rubric tables the score is computed with, so the arithmetic in this matrix reproduces the number exactly.
| Interrupt found | Base weight | × Timing | × Hardness | × Evidence | = Penalty |
|---|---|---|---|---|---|
| Starting score | 100 | ||||
| Mandatory redirect to vendor’s own siteEvery transactionEvery purchase hands off to the merchant's own checkout page to complete (publicly documented product behavior, post-rollback Instant Checkout / Agentic Commerce Protocol model). | 4 | ×8every transaction | ×1soft | ×0.8Documented | −27.2 |
| Total penalty | −27.2 | ||||
| Interrupt Score = 100 − 27.2 = 72.8 → rounds to 73 | 73 | ||||
Checked — nothing found
A check only turns green when a probe tier capable of detecting that interrupt actually ran and found nothing. Not tested ≠ clean.
No checks can be claimed clean yet — no probe tier capable of detecting the remaining interrupt types has run for this listing.
Not yet tested
- CAPTCHA / bot-challengeNo probe tier that can detect this has run yet — unknown, not clean
- Email OTP verificationNo probe tier that can detect this has run yet — unknown, not clean
- SMS / phone OTP verificationNo probe tier that can detect this has run yet — unknown, not clean
- Mandatory phone verification callNo probe tier that can detect this has run yet — unknown, not clean
- Mailed physical code (postal)No probe tier that can detect this has run yet — unknown, not clean
- “Contact sales” wall (no self-serve path)No probe tier that can detect this has run yet — unknown, not clean
- Manual approval queueNo probe tier that can detect this has run yet — unknown, not clean
- ToS clause banning automated/bot accessNo probe tier that can detect this has run yet — unknown, not clean
- SSO-only signup, no API-key issuanceNo probe tier that can detect this has run yet — unknown, not clean
- Rate-limit-triggered manual reviewNo probe tier that can detect this has run yet — unknown, not clean
- One-time account / payment-method provisioningNo probe tier that can detect this has run yet — unknown, not clean
- Identity / KYC verificationNo probe tier that can detect this has run yet — unknown, not clean
- Crypto wallet / key-custody setupNo probe tier that can detect this has run yet — unknown, not clean
Agent-ready alternatives
Other Agent-Commerce Infrastructure tools in the same mode (integrated), ranked by Overall LLM-Friendliness — comparable to this listing because they share its usage mode.