Repository navigation
Settings reference
- Action cards — run generation, generate topics, news topics and AI series suggestions immediately (no waiting for the schedule).
- Usage card — this month's article count, image count, products/categories filled and estimated API cost from real token usage. The monthly counters are fair-use reminders only (5 articles / 6 AI images) — they never block generation. "Show usage history" lists every recorded month plus all-time totals.
- Recent runs — the run log (what ran, what was saved where).
- Queue — pending topics (with move up/down/top, skip, delete), published ones listed separately with a link to each post. Paginated at 10 rows per section.
- Series builder — type a theme and the plugin plans N connected articles as one series (each article links the flow).
- News topics — keywords → Google News RSS headlines added to the queue. "Suggest keywords from my site" drafts niche-relevant keywords with AI.
- AI suggestions — suggested single topics and suggested content series, each addable with one click.
Works when your site has a product post type (e.g. WooCommerce or any custom post type named Product):
- Fill from photo (row action on a product) — AI reads the product image and writes name, descriptions and SEO fields for it.
- Bulk propose (bulk action) — queues many products at once; results are reviewed in the Products tab with preview & approval before anything is written ("apply all" available).
-
Category descriptions (row action on product categories /
product_cat) — AI writes category description + image suggestions.
Daily autopilot toggle, run hour (0–23), active weekdays, target category for new articles, progress mode (background bar vs foreground overlay with auto-refresh).
One active text provider (Claude / OpenAI / Gemini / OpenRouter) + its key; recommended models per provider plus "Custom model…".
Brand name, one-line description of what you do, content language and CTA button text — used in prompts so the writing matches your site. Brand color palette for AI image style.
Image mode (IDs / AI / stock / none), AI image model choice, stock photo provider + API keys (Pexels, Pixabay; Openverse is keyless), optional extra search words appended to stock queries, image optimization toggle, in-body images (0–2), AI image style prompt (placeholder example shown; leave empty for neutral style).
- Detected SEO plugin status (Yoast / Rank Math / SEOPress / AIOSEO) — meta title/description/focus keyword are pushed to whichever is active.
- Internal linking toggle (adds 2–3 internal links per article: products + articles).
- Product links toggle with relevance filter — an article only links a product whose name actually matches the topic; otherwise the CTA points to your contact page/homepage.
- Google Search Console (optional): create an OAuth client in Google Cloud Console, paste Client ID/Secret here, save, then click "Connect" and pick a property. Saved queries become article topics; "Fetch from GSC" pulls fresh query lists on demand.
- Serper.dev key (optional) — powers SERP-based AI topic and series suggestions.
- Own query list — plain text list of your own search phrases used as topics.
Webhook URLs for social sharing (Make.com / Buffer / Zapier compatible). "Test" buttons verify the endpoint from the server side. Optional time slots (3) for automatic product promotion posts; optional AI-written post text. Webhook image fields: image (original), image_square (1080×1080, Instagram), image_pin (2:3, Pinterest).
Keywords for Google News RSS, how many headlines to pull (3–15), "use AI" toggle (AI rewrites headlines into article-ready topic phrasing and checks relevance against your site), optional context note describing what your site is about.
Free-form extra instructions appended to every prompt (tone, length, structure). Outline toggle — plan the outline before writing (recommended). Proofread toggle — a second AI pass polishes the finished text.
Save the current generation settings as a named template; apply or delete templates later. Handy for switching between niches/tones without re-entering everything.
Last generated outline, last image prompt and last raw AI response (truncated) — useful when an article came out wrong: you can see exactly what the model received and returned.