# Discovery engine
For a domain: `robots.txt` (Sitemap: lines, image/video sitemaps skipped, news first, ≤ 8), homepage
`<link rel="alternate">` feeds and links to `status.*`/`*.statuspage.io`, well-known feed paths (`/feed`,
`/rss.xml`, `/atom.xml`, `/changelog/feed`, `/releases.atom`…), well-known sitemap paths, and (if `pages`)
HEAD probes of `/changelog`, `/news`, `/pricing`, `/security`… Every candidate is **fetched and parsed**; only
parseable feeds/sitemaps/statuspages are stored in `discovery_candidates` and promoted (≤ 3 feeds, 1 sitemap,
1 statuspage, pricing/changelog/security pages) according to the source's `discover` flags. Re-run every
`WS_DISCOVERY_INTERVAL_DAYS` (7). CLI: `cli.ts probe <domain>` (dry run), `cli.ts discover [sourceId…]`.