Source to machine record.
Each atlas row becomes a canonical source object with graph role, access class, cadence, entity coverage, and the decisions it can feed.
Pages are no longer the core product. This status layer tracks the source registry, live connector runs, automation risk, and freshness behind the Thailand signal graph.
It separates what exists, what can be automated, what was just checked, and which subject surfaces depend on each source.
Each atlas row becomes a canonical source object with graph role, access class, cadence, entity coverage, and the decisions it can feed.
Live connector probes report success, failure, latency, and sample record counts. Non-polled sources stay visible as registry-only until a connector exists.
Today, stay, route, rail, ferry, safety, and venue decisions should read from this status layer instead of hard-coded source prose.
The accessible candidate set is the practical starting point for Graph Engine v1: public, automation-caveated public, and free-registration sources.
Source access class decides whether a feed belongs in regular automation, cautious verification, paid enrichment, or metadata-only documentation.
These roles tell the truth about actual engine use: which sources are only cataloged, which are merely health checked. And which currently drive observations, signals. And decisions.
Profiled in the atlas, but not yet wired into source health or live graph computation.
311 sources currently sit in this role bucket.
A connector exists and source health probes it, but it does not yet feed the live graph.
10 sources currently sit in this role bucket.
The source currently populates or validates graph entities, but does not yet drive live observations.
0 sources currently sit in this role bucket.
The source feeds normalized observation rows, but not active signal or subject layers yet.
0 sources currently sit in this role bucket.
The source feeds active signal computation, but not traveler-facing decisions yet.
0 sources currently sit in this role bucket.
The source currently influences live traveler-facing decisions through the guide engine.
47 sources currently sit in this role bucket.
Used for standards or explanatory context, not live runtime ingestion.
1 sources currently sit in this role bucket.
These policies are the operating rules for ingestion. They are stricter than page labels because they control whether a source can power live decisions.
Safe target for regular automation, within source cadence and license constraints.
276 sources currently fall into this policy bucket.
Use for low-frequency checks, verification, or only after reviewing automation terms.
12 sources currently fall into this policy bucket.
Use once keys, accounts, quota tracking, and auth status are configured.
20 sources currently fall into this policy bucket.
Use only where paid, partner, or operator access exists.
45 sources currently fall into this policy bucket.
Represent the system or source in the guide, but do not ingest current records.
15 sources currently fall into this policy bucket.
Use for standards, laws, thresholds, and definitions rather than measurements.
1 sources currently fall into this policy bucket.
The static source registry documents what can be checked, what remains gated, and which subject layers depend on each source. It does not claim a live connector run.
Use the static registry for source metadata and the current informations for current status.
The history page explains persisted-run comparisons when that service is connected.
The static build documents source coverage without presenting unavailable live connector data as current.
The registry contains 369 sources with access, cadence, and graph-role metadata.