Llama 4 Scout returns the model output directly as a JSON array in the
'response' field rather than as a plain string or [{"generated_text":"..."}].
The old fallback parsed it as [{"generated_text":""}] (Go JSON ignores
unknown fields) and silently returned an empty string — causing chapter-names
jobs to finish with results:null in the payload despite items_done matching
items_total.
Fix: only accept the generated_text fallback when the value is non-empty;
otherwise return the raw JSON bytes as a string so callers (parseChapterTitlesJSON
etc.) can extract what they need.
Llama 4 Scout returns `result.response` as an array of objects
[{"generated_text":"..."}] instead of a plain string. Decode into
json.RawMessage and try both shapes; fall back to generated_text[0].
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds backend handlers and SvelteKit UI for an admin text generation tool.
The tool lets admins propose and apply AI-generated chapter titles and book
descriptions using Cloudflare Workers AI (12 LLM models, model selector shared
across both tabs).