agentlinkopsHow index status becomes evidence.
An index observation records what a named source reported about one URL at a time. AgentLinkOps keeps the verdict, reason and date together. Link presence, Google index status and search performance remain separate findings.
Google’s URL Inspection API reads its stored index state. It cannot test the live page or request a new crawl through this method. A check here does not submit a URL for indexing.
Five evidence tiers
These are AgentLinkOps evidence labels. Google inspection and a search-result sample carry different confidence, even when both mention the same URL.
| Tier | What supports it |
|---|---|
| Indexed | Google reports PASS with a reviewed positive coverage reason, indexing allowed, a successful recorded fetch and a Google canonical matching the inspected URL. PASS alone is insufficient. |
| Not indexed | Google reports NEUTRAL or FAIL with a nonempty coverage reason that is not one of the accepted positive reasons. The reason remains with the observation. |
| Likely indexed | An exact URL appears in a successfully received licensed site-query sample. This weaker lane has fixtures only; no live supplier requests run. |
| Not found in site query | The received sample lacks the exact URL. It cannot establish that Google has excluded the page. |
| Unknown | A request fails, evidence is missing or incomplete, or the returned fields conflict. A new or unsupported positive coverage reason also stays unknown until reviewed. |
The positive coverage reasons currently accepted are Submitted and indexed, Indexed, not submitted in sitemap, and Indexed. The service asks for English results. A different Google canonical or an unsupported field combination stays unknown; the original response remains available within its retention window.
Google warns that site-query results are not exhaustive. A missing search result never becomes an authoritative negative or confirms an index-loss event. When methods disagree, retain both findings rather than choosing an answer from the weaker sample.
Read the reason and both dates
The check date records when AgentLinkOps obtained the observation. Google’s last crawl time records when Google last crawled the URL. A new check can return the same old crawl date. Google documents these fields in its inspection response schema.
The receipt retains the provider response and its hash, the selected property, canonical URLs and the classified reason. Where the classifier refuses a conflicting result, its reason explains that refusal and the original coverage explanation stays in the inspection fields. A failed request can have no raw response; unavailable evidence is disclosed. After raw evidence expires, the dated receipt remains.
Use the index receipt commands to locate or export saved evidence. Reading a saved receipt does not start another inspection. The data-handling page covers the wider storage boundaries.
When a change is confirmed
The first indexed observation establishes a baseline without a change event. To confirm a later negative state, the current policy requires at least two qualifying Google negatives, at least 30 minutes between counted checks and at least 24 hours from the first negative.
Those intervals are product rules, not Google guarantees. They establish a persistent reported state, not proof of multiple fresh crawls. Unknown attempts do not count toward confirmation and leave the last confirmed state intact. Replayed or out-of-order observations do not advance the count.
An indexed result clears pending negative evidence. An indexed result after a confirmed negative produces a reindexing event. A negative baseline without an earlier indexed result does not produce a deindexing event. These events never overwrite the source link’s own presence or loss state; see the placement confirmation method.
Account access and budgets
The Google lane uses the customer’s connected account and an accessible Search Console property. The watched source URL must fall within that property. Owning the destination site does not grant inspection access to a publisher’s source page. A website owner or administrator must opt in and choose a daily inspection budget.
Google’s URL Inspection quotas include 2,000 requests per day and 600 per minute per property, plus separate Google Cloud project limits. AgentLinkOps enforces its own rolling windows and website budget. Failed or uncertain admitted attempts still consume that budget. Manual and scheduled inspections share it; ordinary link checks use a separate meter.
A cadence is a target subject to quotas and available processing slots. Other tools can also consume the customer’s Google quota, so local counters cannot promise the provider’s remaining allowance. Changing accounts does not turn a property limit into fresh capacity.
What the private pilot has verified
Production Google inspection and scheduling are on for websites that opt in; staging remains off. On September 20, 2026, the first scheduled check through a connected customer account returned an indexed observation for our own how-it-works page. The retained response reported Submitted and indexed, with a September 15 crawl date. This was an owned internal-navigation watch, not an earned backlink.
Repeated scheduled checks and reason history over time still need verification. The licensed site-query lane remains fixture-only. No index verdict here establishes rankings, traffic or the effect of a backlink.
Read the dated capability record for availability and the monitoring workflow for the separate placement checks. Reviewed September 20, 2026 against the inspection classifier, quota admission and retained first scheduled receipt.