Exactly why your server did not meet a condition.
Not the generic answer. The bytes we actually got back from your endpoint, the condition that did not hold, and what to change in your code or your proxy. Then we measure it again after you have changed it.
That is a conflict of interest, and naming it is not a defence. So here is what was done about it, in the machine rather than in this paragraph.
The verdict is produced by a machine and there is no code path that takes a payment and changes a status. Not buying this changes nothing about your row. Buying it changes nothing about your row either. Only the next measurement does.
The free path is complete, not a teaser. What to change for each of the five conditions is published in full on the operators page, and asking for a re-measurement is free forever. Most operators should stop there and never come here.
The price is arithmetic you can check. Hours times a published rate, with the hours itemised below. It is not a number chosen by looking at how badly you want the row to be green.
Three free things that solve most of it
Read what the condition actually measures
Each of the five is a specific request and a specific field, written out. Most rows that are not green are one missing key in one JSON file. The five conditions.
Recompute our verdict yourself
If our hash does not reproduce, the defect is ours and fixing it is our job, not something you should be paying for. Recompute.
Ask for a re-measurement
Free, no account, and you do not have to own the server. If you already changed something, this is all you need. The corrections desk.
The measurement, opened up
Hours times a rate, and both are printed
| Work | Hours |
|---|---|
| Measure the endpoint and read the actual bytes it returned | 0.5 h |
| Identify the cause per condition and write what to change | 1.0 h |
| Re-measure after your change, and confirm | 0.5 h |
| One endpoint | 2.0 h |
| Each further endpoint on the same host | 1.0 h |
Can an agent tell your failure from your silence?
The five conditions above are pass or fail. Condition 06 is not. It is a disclosed measurement, free and with no verdict: from tools/list alone, does each tool's declared outputSchema let a consumer separate “the call failed” from “the call found nothing”? Across the public field, 96.7 to 99.1% of tools cannot, and the field names behind every score are published so you can see the false positives yourself. The free measurement is on the operators page.
This is the paid work that changes your number, and it is engineering, not a mark. A real read-state where none exists (a boolean, or an enum such as lookup: ok | absent); failures moved into the tool-error channel (isError) instead of riding home as a successful-looking result; the correct shape carried across every tool rather than one and not its neighbour. We measure your number before and after, under one ruler, in public, with the gate as the witness.
85.7% of our own tools could not tell failure from silence, the verification gate and the Bitcoin-anchored ledger included, every tool on each. That is slightly better than the field and still almost all of ours: building the checker did not exempt us. We fixed it to 2.9%, derived the before-figure from the pre-change files so the two are comparable, and froze the whole thing, the assumption, the four live bugs under it, and the reproduce commands, in our public append-only ledger as record mould-gh-12. The seller took the medicine first, in public.
The same rails apply as everywhere here. Condition 06 is a measurement, not a pass, so there is nothing to buy that changes a verdict. We do not send this offer to the 12,429 endpoints we measured; you came to the free number, and the offer is here for when you want the next step. And a schema is a declaration, not behaviour: we fix the declared contract and the runtime path behind it, and where only tools/call would prove the runtime, we say so rather than claim it.
Stated at the same size as the price
- ✗Not a passing verdict. There is no mechanism. The next measurement decides, and it does not know whether you paid.
- ✗Not removal of the row you have now. The history stays, including the stretch where it was not green. Ours does too.
- ✗Not priority. There is no faster queue, because there is no queue. Measurement is automatic and free for everybody at the same speed.
- ✗Not silence. Your row says what it says whether you buy this or not.
Send the endpoint
Write to contact@the-horizons-innovation.com with the endpoint and, if you know it, which condition you are stuck on. If we look and think you can fix it from the free pages in ten minutes, we will tell you that instead of taking the work.
条件を満たさなかった原因を、御社のサーバーについて個別に調べて返します。返ってきた実際の応答、満たさなかった条件、直す箇所、そして直したあとの再測定まで。
1本 2.0時間 44,000円（税込）。同じホスト上の2本目以降は1本あたり 1.0時間 22,000円（税込）。単価 22,000円／時（税込）は接続前審査と同じです。請求は納品後です。無料のページから得られる以上のものが無かったと言われた場合、請求しません。
ただし、その前に無料の道が3つあります。条件の定義、判定の再計算、再測定の依頼。ほとんどの場合はそれで足ります。足りたなら、ここに来る必要はありません。
そして、お金を払っても判定は動きません。支払いを受けて状態を変える経路が、コードの中にありません。