Measured 2026-09-16 · public surface only
An assistant stops at: discover it → buy without a human → pay as a machine
Verdict: halfway — 4 failing checks out of 11 we run.
| Gate | Result | Evidence |
|---|---|---|
| discover it | an agent stops here | missing (HTTP 200) |
| read the catalog | passes | present: SoftwareApplication |
| get past robots.txt | passes | no usable robots.txt (HTTP 200) — nothing is blocked |
| find a machine price | passes | found at https://sparkpost.com/pricing |
| buy without a human | also fails | purchase path gated by: Sign up, Contact sales |
| pay as a machine | also fails | /api returned HTTP 406, no 402 challenge |
| Failing check | What we measured | The fix |
|---|---|---|
| llms.txt present | missing (HTTP 200) | Publish `llms.txt` — the map an agent reads first. |
| sitemap.xml present | 0 URLs | Publish `sitemap.xml` so an agent can enumerate what exists. |
| Purchase possible with no human signup | purchase path gated by: Sign up, Contact sales | Add a paid path that needs no account — an HTTP 402 challenge the agent can settle. |
| Machine-native payment challenge (HTTP 402) | /api returned HTTP 406, no 402 challenge | Return HTTP 402 with a payment challenge instead of a signup wall — this is the one nobody has. |
We audited 206 companies the way an assistant reads them. None can accept payment from a machine, and this one is no exception: the documentation checks are widely passed, the payment checks are almost universally failed. Discovery earns nothing on its own — the money is in the last gate.
The full report, method and raw data →
Free, ten seconds, no signup — and the score is public either way.
llms.txt, robots.txt, sitemap.xml, schema.org markup, /.well-known/*, the pricing page, and an HTTP 402 probe.node audit.js sparkpost.com.