Use this master guide to understand the complete Publisher 3.7.6 system before configuring an individual screen.

Who this guide is for

Start here if you are installing Publisher, taking over an existing configuration or auditing an automated publishing operation. The detail guides linked below contain every field and recovery procedure; this page explains how the parts work together.

The complete workflow

1. Licence and providers. Publisher confirms that the licence grants runtime access and that the selected text provider has a usable key. Image generation is optional, but requires its own supported provider and key.
2. Inventory. Publisher reads public, publicly queryable WordPress content that is not excluded from search. Posts, pages, products and eligible custom post types can supply topic evidence and internal-link candidates. Attachments are excluded.
3. Strategy and topics. Website inference uses titles, content, headings, excerpts, public taxonomies, SEO keywords and product information to propose clusters and topics. You choose the language, goal, review mode and minimum score.
4. Research. A job checks overlap against the inventory, prepares internal-link candidates and, when configured, adds Serper and Search Console evidence. These integrations enrich research but are not required.
5. Brief and structure. Publisher chooses an article archetype and a varied set of content modules, then builds a brief and a section plan appropriate to the topic, intent and requested length.
6. Section writing. The article is generated in checkpoints rather than as one opaque response. This supports live progress, retry and recovery without discarding all completed work.
7. Links and sources. Internal links must point to real allowed URLs and are selected for relevance in the same language. External citations must be verifiable. A relevant-link shortage is reported instead of being filled with unrelated links.
8. Images. When enabled, Publisher creates a featured image and a section-aware inline plan. It varies the visual treatment and paragraph position, avoids unsuitable sections, stores WebP media and writes alt text.
9. Quality gate. Publisher checks length, structure, headings, required modules, links, sources, media, unresolved markers, repetition and other safety conditions. Blockers prevent automatic publication; non-blocking shortages remain visible as warnings.
10. WordPress output. The generated item is a native post, even when custom post types contributed inventory evidence. Publisher assigns an eligible author, categories, tags, comments, language and SEO-plugin metadata.
11. Schedule or review. The final result follows the selected draft, pending-review or auto-publish workflow. Automatic publication requires a quality-ready article and a valid schedule slot.

What you control and what Publisher automates

Required and optional components

• Required: WordPress 6.0 or newer, PHP 7.4 or newer, an active Publisher licence and one configured text provider.
• Required for AI images: OpenAI or Gemini image configuration. Text-provider tests do not prove that image generation works.
• Optional: Serper, Google Search Console, an SEO plugin, WPML or Polylang, Action Scheduler and a server cron. Publisher has documented fallbacks when these are absent.

A safe first-time configuration

1. Back up the site, install Publisher and activate the licence for the correct environment.
2. Configure one economical supported text model. Leave image generation off until the first text-only job works.
3. Keep the final post status as Draft, select a real eligible author and use WordPress defaults for comments.
4. Set a realistic length range, keep the quality threshold at 82 and use the default module range.
5. Build the inventory, inspect System Status and generate a small topic batch in Review mode.
6. Approve one narrow topic and create one Draft from Create Articles.
7. Inspect Live Pipeline, the WordPress post, article report, links, categories and logs.
8. Enable images, optional research integrations and scheduling one at a time. Verify one controlled job after each change.

How to know the system works

A successful test is not merely a completed job. Confirm that the job reached the expected final state, the article opens as a native WordPress post, its author and language are valid, every link resolves, media appears in the library with alt text, the report distinguishes requested and actual counts, the quality gate passed and the schedule/history record matches the intended workflow.

Failure boundaries and graceful fallback

• No Search Console or Serper: research continues with inventory and model context; the report records the missing optional source.
• Too few relevant links: Publisher warns and continues rather than inserting unrelated destinations.
• Temporary provider or worker failure: the job resumes from checkpoints after backoff when the error is retryable.
• Quality blocker: the article remains non-published for correction; lowering the threshold should not be the first response.
• Missed schedule slot: the configured policy can use the next ready article or skip the slot; it does not publish an unready article.
• Invalid licence: runtime actions stop and Publisher screens become read-only until access is restored.

Detailed guides

• Install, activate and create the first article
• Dashboard, System Status and Content Inventory
• AI text and image providers
• Topic Manager and Smart Topic Strategy
• Brand voice, instructions, length and structure
• Inventory and cannibalisation control
• SEO, links, citations and sitemaps
• Images, alt text and WebP
• Automation and scheduling
• Create Articles and Live Pipeline
• Jobs, logs and recovery
• WordPress-native articles and attributes
• Quality rules and reports
• Search Console, Serper and multilingual sites
• Install, update, migrate and roll back safely

Master checklist

• Licence active for the correct domain and environment.
• Selected provider has the matching saved key and usable model.
• Inventory and sitemap status are current.
• Topic review mode and minimum score match the editorial policy.
• Length, voice, link, image and quality settings reflect the site.
• One manual draft has passed end-to-end verification.
• Background processing completes reliably before automation is enabled.
• Only quality-ready content can enter an automatic publication slot.

Publisher 3.7.6 native scheduling and reservation

After quality approval, Draft-only keeps an editable Draft and Review keeps a Pending post. Auto publish reserves the article to exactly one future publication slot. When the slot is more than 60 seconds away, Publisher sets the native WordPress status to Scheduled (future) and writes the exact date in the WordPress site timezone.

The reservation prevents another slot from reusing the article. WordPress’s future-post event is the normal publication path. Publisher keeps a separate watchdog event that verifies the native result and can safely recover a missed slot according to the next-ready/skip policy.

Durable lifecycle

The automation ledger can show planned, queued, generating, generated, ready, scheduled, review, publishing, published, skipped, failed or cancelled. Generated, Scheduled and Published timestamps connect the content job, WordPress post and intended slot.

3.7.6 verification

• Auto-publish content appeared as a native Scheduled post before the slot.
• The post date matched the WordPress site timezone.
• The article was reserved to one slot only.
• WordPress publication and Publisher history both reached Published.
• Pausing or switching to Review/Draft-only removed the future schedule without deleting the article.

Missed publication slot: inspect the Scheduled post and site timezone, then WP-Cron/Action Scheduler and Publisher history. Use next ready may select another unreserved quality-approved article; Skip records the miss. Neither policy publishes incomplete, blocked or already-reserved content.