Manifest
The build-time record of all content, with entries, public views, lookup maps, the content graph, and diagnostics. Pages, runtime queries, incremental builds, and inspection all read from the manifest.
What the manifest holds
| Field | Contents |
|---|---|
entries |
Every processed Content Entry |
publicEntries |
Entries safe to publish |
discoverableEntries |
Entries allowed in listings and feeds |
bySlug, byPermalink, byAlias |
Lookup maps |
graph |
The content graph |
diagnostics |
Collected warnings and errors |
Manifest Lookup Maps
The ContentManifest provides several specialized views for different access patterns:
| View | Contents | Typical use |
|---|---|---|
bySlug |
Map<string, ContentManifestEntry> |
Look up an entry by its slug. |
byPermalink |
Map<string, ContentManifestEntry> |
Look up an entry by its permalink URL. |
byAlias |
Map<string, ContentManifestEntry[]> |
Find entries by their alias/redirect paths. |
byRoutablePermalink |
Map<string, ContentManifestEntry> |
Look up only routable entries (public + unlisted, excluding draft/scheduled). |
byTag |
Map<string, ContentManifestEntry[]> |
Group entries by tag for taxonomy pages. |
byAsset |
Map<string, ContentManifestEntry[]> |
Group entries by asset for asset manifest generation. |
publicRedirects |
Map<string, ContentRedirect & { slug: string }> |
Public redirects with owning entry slug. |
These views are generated during the build and are pure projections — they read entry data from the manifest without requiring any filesystem access. For example, to find an entry by permalink:
const entry = manifest.byPermalink.get("/blog/my-post/");Graph and Diagnostics
The graph field holds a ContentGraph instance, enabling queries such as graph.outgoing(slug), graph.incoming(slug), and graph.neighbors(slug). This is used by backlink displays, related-links sections, and internal link validation.
The diagnostics field is an array of Diagnostic objects collected during content processing. Each diagnostic includes a level (info, warn, error), source, message, and range (source position). Downstream consumers (page generation, runtime queries, inspection) read from the manifest exclusively — nothing downstream needs to touch the filesystem.
The boundary between stages
Content processing produces the manifest; page generation, runtime queries, incremental builds, and inspection consume it. Nothing downstream needs to touch the filesystem.