Public data and evidence guide

The text discovery files summarize publisher identity, answer boundaries, and important routes. The JSON discovery document provides structured categories and citation guidance. The source feed publishes code-owned regulatory references with computed age and freshness signals. Sitemap and robots files serve crawling decisions but are not content databases.

These endpoints are intentionally limited to public data. A field appearing in a discovery document means that the publisher chose to expose that description; it does not prove real-time operator availability, account eligibility, a current market price, or a payment result. Consumers should follow the canonical page and its source before using a field in a consequential answer.

A source record identifies the named organization, topic, URL, and repository review date. Age values are computed from that reviewed date to help prioritize verification. A fresh flag means the record is inside the configured review window; it does not certify that every statement on the external page is unchanged or that an editor has approved a new conclusion.

Canonical and indexability decisions come from the curated content manifest. Routes awaiting review, redirects, merged pages, and noindex support pages stay out of public discovery. The manifest deliberately avoids manufacturing last-modified dates from builds or saves, because synthetic freshness would give crawlers and readers a misleading editorial signal.

Public discovery responses must not contain email addresses from user submissions, account identifiers, click histories, private prompts, access tokens, internal service locations, or secret-gated workflow routes. Analytics and operational records have separate consent, authorization, and retention boundaries and are not joined into these public files.

The public documents are designed for discovery, not as a guaranteed versioned API or bulk data product. Field shape, cache policy, and route selection can change when the editorial contract changes. Consumers should tolerate missing optional fields, respect canonical URLs, retain source attribution, and avoid caching time-sensitive conclusions beyond the underlying evidence window.