Open discovery · 0/9
ChatGPT, Claude, and Gemini did not return Signmos in any of the nine frozen open-web runs.
Verification package
Signmos reference case
One product, documented in stages. This navigation keeps the current stage and its published artifacts together.
The measurement report expires. The summary, consumer-agent results, and public JSON digests remain as the durable record.
Signmos before changes.The first public record.
We recorded the public state of Signmos before making changes. Auditmos operates Signmos and this test. This checks the method; it is not independent proof of demand.
See the resultPlaces checked11
Available4
Missing7
HTTP differences0
Result
Signmos was absent from all nine open discovery answers.
Named vendor · 0 qualified
Eight of nine evaluations ended with Insufficient Evidence. One Gemini run evaluated the unrelated SignsOS product and returned Disqualified.
Source retrieval varied
ChatGPT used relevant sources in three of three verification runs. Claude found none; Gemini ranged from no source to a wrong vendor to relevant Signmos sources.
What it means
The consumer baseline supports the information-gap hypothesis.
For this exact task and run window, the public surface did not let the sampled consumer products qualify Signmos confidently. The technical files said more than the buyer page, but the products did not retrieve them consistently.
What it does not prove
A missing answer is not an explanation of the cause.
The 0/9 result records these consumer answers. It does not prove provider index state, ingestion, ranking cause, or what every user and later model version will receive.
Correction
We did not confirm the earlier blocking result.
In the repeated test, the available resources responded in all seven checks. We have no evidence that Signmos broadly blocks buyer agents.
See the full record, results, and comparison plan
Public evidence record
Two public records preserve the surface and consumer results.
This page is the permanent record for people. One JSON file preserves the website measurement; the other preserves the consumer scope, run counts, verdicts, grading totals, evidence boundary, and links to raw records.
- Record
- Signmos S0 / 20260829T094225Z
- Captured
- 29 Aug 2026 · 09:42:25 UTC
- Observer
- Website check 0.4.0 + Browser Run · Cloudflare · WAW
- Method
- surface-snapshot/0.4.0 + visual-capture/0.2.0
- Evidence ceiling
- Reachable
- Ownership
- Auditmos operates this program and Signmos
Download the public JSON record Download the consumer-agent JSON record Open the page-by-page report
The service report and its rendered images stay online until 28 Sep 2026. This page and the public JSON digest remain the permanent record.
S0 work status
The consumer baseline is complete. The comparison can now start.
The site capture and 27 consumer-agent runs are recorded. A later change still proves nothing until the same consumer tests are repeated.
- 01Complete
S0 public pages recorded
We recorded eleven resource concepts, including bounded buyer-page searches, before changing Signmos.
- 02Complete
Result and correction published
The latest 0.4 run recorded four reachable and seven missing concepts plus six rendered views.
- 03Complete
Consumer buyer-agent tests
We saved 27 full runs from ChatGPT, Claude, and Gemini. The controlled API task is explicitly deferred outside the consumer scope.
- 04Ready
Make the suggested changes
The before-state is now fixed. Publish the missing buyer evidence without changing the frozen prompts.
- 05Waiting
Repeat after deployment
We will publish a result only after we confirm the changed version and repeat the same test.
Before → recommendation → after
Keep the starting record. Do not replace it with a promise.
The left side shows what S0 found. The middle lists the suggested change. The right side stays Pending until we deploy the change and repeat the same test.
| Area | S0 · observed | Suggested change | After the change · measured |
|---|---|---|---|
| AreaBuyer information | S0 · observedNone of four bounded searches found a reachable buyer page. They tried 10 pricing, 7 security, 8 privacy, and 8 terms candidates. | Suggested changePublish standard web pages for the offer, price, security limits, privacy, and terms. | After the change · measuredPending - we have not measured a Signmos change. |
| AreaDiscovery | S0 · observedrobots.txt worked. sitemap.xml was missing. The homepage exposed no same-origin buyer-page links in the captured metadata. | Suggested changeAdd a sitemap, canonical URLs, accurate social data, and internal links before testing agent files. | After the change · measuredPending - confirm page access and outside platform requests separately. |
| AreaProduct fit | S0 · observedThe homepage gave four tasks equal weight. It did not name one target customer, a price, or key limits. | Suggested changeLead with one buyer need, honest limits, the operator's identity, and one clear next step. | After the change · measuredPending - repeat the same vendor test. |
| AreaOperational evidence | S0 · observedagent.md and OpenAPI had more detail than the buyer pages. | Suggested changeLink technical facts to clear buyer pages. Do not treat a working API as proof that the product is ready to buy. | After the change · measuredPending - test whether answers cite more supported facts. |
| AreaAgent behavior | S0 · observedSignmos appeared in 0 of 9 open discovery runs. Named-vendor evaluation returned 0 Qualified, 8 Insufficient Evidence, and 1 wrong-vendor Disqualified result. | Suggested changePublish complete buyer evidence, then repeat the same three consumer scenarios in the same products. | After the change · measuredPending - no post-change consumer run exists yet. |
Still unknown
The consumer result is recorded; the cause remains unknown.
- Whether each provider indexed, cached, or ingested the Signmos pages.
- Why a platform missed Signmos or selected an unrelated vendor in a particular run.
- Outside-platform request logs and the exact Signmos production commit.
- Whether the same answers recur after the public buyer surface changes.
Connected article
The explanation and the evidence link to each other.The article explains why buyer agents need a chain of evidence. This case now includes the public-surface record and 27 consumer-agent runs. The protocol preserves the exact questions and grading rules.