@koda-sl/baker-cli 0.353.0-dev.afbf521d7 → 0.353.0-dev.df03f88fe

package/README.md

@@ -2536,6 +2536,16 @@ baker images extract https://example.com --wait-for 1500 --limit 20
 
 Every image extracted from a page shares the scraped-page URL as `externalUrl`. Big pages can return 100+ images — cap auto-ingest at 20.
 
+A failed extract says which kind of failure it was, with `retryable` and a `fix`:
+
+| Code | Meaning | Next move |
+|------|---------|-----------|
+| `UNAVAILABLE` | The page reader did not answer or timed out just now | Retry once, then fall back |
+| `VALIDATION_ERROR` | The page itself refused (e.g. HTTP 404) or is too large to read | Do not retry; fall back |
+| `INTERNAL_ERROR` | A fault on Baker's side | Do not retry; fall back |
+
+The fallback is `baker images screenshot <url>` for the page as one image, or `baker images logo <domain>` for the brand's mark. A `503` from any command now maps to `UNAVAILABLE` instead of `INTERNAL_ERROR`.
+
 **Flags:**
 
 | Flag             | Description                                                                                |
@@ -3875,17 +3885,12 @@ baker canvas validate my-canvas.json
 #    or the same ad rendered once per aspect ratio instead of one hero + a terminal
 #    image_aspect_adapt (IMAGE_PARALLEL_FORMAT_RENDERS — separate renders drift in type,
 #    crop, and color, so the sizes stop matching).
-#    A generator prompt is measured with its {{slots}} filled from text nodes and local
-#    prompt.json/text ingests: past the model's character cap it is refused as
-#    PROMPT_OVER_MODEL_LIMIT, naming the node and the overshoot, before anything bills.
 
 # 2. Run (executes nodes, writes files to ./canvas/<run_id>/)
 baker canvas run my-canvas.json
 #    Advisory warnings print as `[warn] <CODE>: …` before anything bills.
 #    Only nodes reachable from the declared output execute (orphaned nodes are skipped,
 #    not billed). Add --keep-runs N to prune old r_* run dirs, keeping the N newest.
-#    A failed run reports `error.code: "input_invalid"` when the backend refused a node's
-#    own input (it fails the same way until the input changes), `"runtime"` otherwise.
 
 # 2b. Independent nodes ALREADY fan out in parallel — every node in the same
 #     topological layer (e.g. all the per-clip video_generate nodes) runs concurrently