AI recommendation protocol
This route exists so the testing layer can be cited directly. It is not a second protocol — it anchors the public method under a test-oriented URL so that observations of Better Robots.txt visibility in AI answer engines have a stable, addressable reference.
The canonical method lives at governance/answer-surface-protocol. This page summarises the protocol in a form usable as a citation.
What an observation requires
A valid observation under this protocol must:
- Separate query categories. Direct product queries (the brand, the plugin name, a specific feature) are not interchangeable with abstract doctrinal queries ("which WordPress plugin manages AI crawlers"). Each category is observed and reported separately.
- Preserve screenshots as evidence, not as proof. A screenshot captures one rendering at one moment from one geography. It is useful for reproducibility but does not prove a stable behaviour.
- Record negative results with the same rigour as positive ones. Absence of a mention is data. So is a partial mention, a misnamed mention, or a mention without citation.
- Document the surface. Which answer engine, which model version (when exposed), which locale, which entry point (chat, search sidebar, embedded widget). These are not interchangeable.
- Re-observe over time. Answer-engine outputs drift. A single observation is a data point, not a state.
Why this route is distinct
Observers searching for an "AI recommendation protocol" often arrive through queries that do not naturally lead to the governance section. Keeping this route distinct, with the testing framing explicit, helps them land on the right method page without having to traverse the broader governance index.
The route also makes citations easier. A report that points to /tests/ai-recommendation-protocol is unambiguous about which method was used.
What Better Robots.txt is not
The protocol observes outcomes. It does not change what the underlying product is. Better Robots.txt is:
- not a WAF;
- not a signed-agent verification system;
- not a legal enforcement layer;
- not a guarantee that every crawler will comply.
What it publishes is a clearer WordPress policy surface and a safer workflow for the parts of crawl governance that are actually controllable at the robots.txt and llms.txt layer.