scribe-cms 0.0.12 → 0.0.14

package/README.md

@@ -4,7 +4,7 @@ Typed, file-based CMS for multilingual MDX. English source files on disk, locale
 
 Scribe has no framework dependency — it reads files and SQLite in-process and works with any Node-based stack (Next.js, Astro, Remix, SvelteKit, a static-site script, …). Examples in these docs use Next.js, but nothing about Scribe is Next-specific.
 
-**Docs:** [Getting started](./docs/getting-started.md) · [Configuration](./docs/configuration.md) · [Writing content](./docs/content.md) · [Runtime API](./docs/runtime-api.md) · [Translation](./docs/translation.md)
+**Docs:** [scribe.genlook.app/docs](https://scribe.genlook.app/docs) · [Getting started](./docs/getting-started.md) · [Configuration](./docs/configuration.md) · [Writing content](./docs/content.md) · [Runtime API](./docs/runtime-api.md) · [Translation](./docs/translation.md)
 
 ## Install
 
@@ -40,6 +40,7 @@ export default defineConfig({
   // store: ".scribe/store.sqlite"     (default)
   locales: ["en", "fr"],
   // defaultLocale: "en"              (default)
+  // localeFallbacks: true            (default: regional variants fall back to their base language, e.g. pt-BR to pt; set false to disable)
   types: [
     defineContentType({
       id: "blog",
@@ -82,7 +83,7 @@ const scribe = createScribe(config);
 // Lists & lookups
 scribe.blog.list("fr");                       // sorted docs for a locale
 scribe.blog.get("my-post");                   // exact slug lookup, no fallback
-const r = scribe.blog.resolve("my-post", "fr"); // cross-locale slug fix + EN fallback
+const r = scribe.blog.resolve("my-post", "fr"); // cross-locale slug fix + fallback chain + EN fallback
 // r = { document, actualLocale, shouldRedirectTo?, canonicalPath? }
 
 // Routing helpers
@@ -105,13 +106,27 @@ are inferred from the config — no codegen.
 
 ```bash
 scribe status                  # EN docs + translation coverage
-scribe validate                # schemas, relations, redirects, sqlite consistency
-scribe translate --locale fr   # translate stale/missing pages (Gemini)
-scribe translate --preset active
-scribe history blog my-post fr # revision timeline
+scribe validate                # schemas, MDX bodies, relations, redirects, sqlite consistency
+scribe translate --locale fr   # translate stale/missing pages (Gemini Batch API, 50% token cost)
+scribe translate --preset active --strategy missing-only
+scribe translate --direct      # per-page API calls at full price, immediate results
+scribe translate --resume      # pick up pending batch jobs from an interrupted run
+scribe translate --dry-run     # show the worklist without calling the API
+scribe history blog my-post fr # EN snapshot timeline for one document
 scribe studio                  # read-only local admin UI
+scribe export-static           # raw MDX files for static hosting / crawlers
 ```
 
+Translated output is checked before it is stored: the returned MDX body must
+parse and the frontmatter must re-validate against your Zod schema, so a bad
+model response fails the command instead of reaching production. Interactive
+runs show live progress with token counts and an estimated cost in USD.
+
+Translation runs through the Gemini Batch API by default (half the token
+cost). Jobs are persisted in the store as soon as they are submitted, so
+interrupting a run is safe: the next `scribe translate` (or `--resume`) picks
+pending jobs back up without submitting anything twice.
+
 Translations are stored in `.scribe/store.sqlite` keyed by a hash of the EN
 translatable content, so `scribe translate` only re-translates what changed.
 **Commit `.scribe/`** — do not add it to `.gitignore`.