Glossary
The core WordPresto vocabulary, grouped by kind. These are the canonical definitions the product, the assistant and these docs all share.
Concepts
Section titled “Concepts”Project
Section titled “Project”A workspace WordPresto builds intelligence around — either a website or a single piece of writing.
A Project is the unit of work in WordPresto. There are two kinds: a Website Project (a living intelligence model of a domain) and a Writing Project (a living model of a single writing task). Both use the same central Canvas and Worker system. A project must be a persisted record with a real id before project-scoped features run; a temporary domain:example.com intent key is not a project id.
Related: Website Project, Writing Project, Project Intelligence
Website Project
Section titled “Website Project”A living intelligence model of a domain — its pages, search data, structure and opportunities.
A Website Project progressively learns a domain: discovered pages, titles and metadata, headings, canonical URLs, indexability, internal links and site architecture, plus Search Console, GA4 and Bing metrics, keyword and SERP enrichment (after an explicit action), crawler findings, keyword sets, the topical map, content gaps, cannibalisation candidates and internal-link opportunities. That knowledge is the Project Intelligence that powers Emma, the Workers and Visual Intelligence.
Related: Project, Project Intelligence, Crawler, Site Health
Writing Project
Section titled “Writing Project”A living intelligence model of a single writing task — its audience, intent, format and constraints.
A Writing Project knows what the user is writing and for whom: audience, intent, format, destination, tone and voice requirements, source material, editorial constraints and approval requirements. It uses the same Canvas and Workers as a Website Project but does not need crawler or domain intelligence.
Project Intelligence
Section titled “Project Intelligence”The persisted, source-labelled, project-scoped source of truth that powers every WordPresto surface.
Project Intelligence is the shared knowledge layer behind Visual Intelligence, Project Home, Emma, the Workers and the Canvas. Every datapoint is persisted (never client-only), labelled with its source (Search Console, GA4, Bing, keyword enrichment, crawler, page analysis, keyword set, CMS, learning, Canvas decision or worker output) and scoped to a project. Nothing here is invented — a datapoint only exists because a real source produced it.
Related: Project Intelligence Datapoint, Visual Intelligence, Emma
Project Intelligence Datapoint
Section titled “Project Intelligence Datapoint”One persisted, source-labelled fact about a project.
A datapoint is a single record in the Project Intelligence layer. It carries an id, project id, source and source type, entity type, a label, the relevant URL/query/keyword, optional metrics and relationship JSON, provenance, a confidence and timestamps. Datapoints are never fabricated: an empty result means the source was queried and returned nothing, not that a value was invented.
Related: Project Intelligence, Google Search Console, Crawler
Canvas
Section titled “Canvas”The central rich editor where content is created, reviewed by Workers and approved.
The Canvas is WordPresto’s main work surface — a rich editing environment, not a textarea. Content is created, imported, reconstructed from a live page, edited, proofed, rewritten, reviewed, compared, versioned, approved and prepared for destination handoff here. Its content model is CMS-agnostic so it can map to Astro, Sanity, WordPress and others. The Canvas is the source of approved content; editing approved content returns it to draft/review.
Related: Worker, Approval, Builder Worker
WordPresto’s assistant — she understands the active project and routes work into the Canvas.
Emma is not decorative chat. She understands the active project, its type and data sources, the current Canvas state, recent worker outputs, approval status and what setup is missing, and she routes content work into the Canvas. She answers from real Project Intelligence and the shipped help guides and glossary rather than guessing, and she can hand off to human support when needed.
Related: Project Intelligence, Canvas
Visual Intelligence
Section titled “Visual Intelligence”The visible map of a project’s real, persisted intelligence.
Visual Intelligence renders the Project Intelligence layer: source counts, query/page/keyword/URL nodes, and the relationships, gaps and opportunities that are genuinely derived from real data. It never invents nodes, links, clusters, gaps or cannibalisation, and it never shows zero datapoints when real persisted data exists.
Related: Project Intelligence, Topical Map
Topical Map
Section titled “Topical Map”The structure of topics and clusters a project genuinely covers, derived from real data.
A Topical Map organises a project’s topics and clusters. It is derived from real Project Intelligence (queries, pages, keywords and their relationships), never invented, and feeds content-gap and internal-link analysis.
Related: Content Gap, Keyword Set
Content Gap
Section titled “Content Gap”A topic the data shows a project should cover but does not yet.
A Content Gap is a genuine coverage hole, derived from real search and topical data rather than guessed. It points to content worth creating and is surfaced only when the underlying data supports it.
Related: Topical Map, Keyword Set
Cannibalisation
Section titled “Cannibalisation”When two or more pages compete for the same query, splitting their performance.
Cannibalisation is when multiple pages on a site target the same query and undercut each other in search. WordPresto surfaces candidates only from real query-to-page data; it does not fabricate findings.
Related: Internal Link Opportunity, Topical Map
Internal Link Opportunity
Section titled “Internal Link Opportunity”A specific, data-backed chance to link one page to another.
An Internal Link Opportunity is a concrete suggestion to link between two of a project’s pages, derived from real page and topical relationships. It supports site structure and helps resolve cannibalisation, and is only shown when the data supports it.
Related: Cannibalisation, Topical Map
Site Health
Section titled “Site Health”A score and explorer for the technical state of a crawled site.
Site Health measures the technical state of a site from real crawler findings — status codes, titles, meta descriptions, canonical URLs, indexability and structure. The score is explained honestly (what it measures and what it excludes, such as attachment pages and archives), with an All Pages explorer and a refresh path.
Related: Crawler, Project Intelligence Datapoint
Keyword Set
Section titled “Keyword Set”A named group of keywords a project is working with.
A Keyword Set is a persisted group of keywords scoped to a project, used to focus research, the topical map and content planning. Keywords come from real research; paid keyword enrichment only runs on an explicit action.
Related: Keyword & SERP enrichment, Topical Map
Learning
Section titled “Learning”The record of editorial decisions — accepted, rejected and edited suggestions — that teaches WordPresto a project’s preferences.
Learning captures the user’s editorial decisions: accepted and rejected suggestions, manual edits, preferred phrasing, style rules and recurring corrections. It is distinct from provider and crawler facts (which live in Project Intelligence); learning records preferences, not source-of-truth metrics.
Related: Worker, Project Intelligence
Workers
Section titled “Workers”Worker
Section titled “Worker”A specialist agent that reads the project and Canvas, does one task, and returns suggestions for human review.
Workers are specialist agents that operate around the Canvas and Project Intelligence. A Worker reads the active intelligence and current Canvas content, performs one specialist task (drafting, rewriting, proofing, SEO titles, schema, evidence gaps, accessibility, internal links, cannibalisation and more), and returns structured suggestions. Workers never write to external systems silently, never fabricate data and never bypass Canvas approval; accepting, rejecting or editing their output feeds Learning.
Related: Canvas, Builder Worker, Learning
Builder Worker
Section titled “Builder Worker”The Worker that packages approved Canvas content for a chosen destination.
The Builder Worker prepares approved Canvas content for a selected destination, taking the content plus formatting instructions, platform, page/content type, metadata, schema, media, slug and internal-link requirements and CMS field mapping. It outputs a destination-ready package and never invents destination fields — if schema or mapping is missing it returns destination_schema_required or mapping_required.
Related: Canvas, Publishing Agent, Approval
Publishing Agent
Section titled “Publishing Agent”A destination-specific agent that hands approved, built content to a CMS such as Astro or Sanity.
A Publishing Agent takes the Builder Worker’s package and hands it to a specific destination — for example the Astro agent (content collections, Markdown/MDX frontmatter) or the Sanity agent (document type, Portable Text, mapped fields). It validates destination requirements, prepares a preview or draft by default, and only publishes live on an explicit action. It never publishes from an unapproved Canvas and never fakes a successful publish.
Related: Builder Worker, Approval
Data sources
Section titled “Data sources”Google Search Console
Section titled “Google Search Console”Google’s search-performance source — queries, pages, clicks, impressions, CTR and position.
Search Console is a Google data source WordPresto ingests once the account is connected and a site is mapped to the project. It contributes queries, pages, query-to-page relationships, clicks, impressions, CTR and average position as source-labelled datapoints. If connected but unmapped it shows google_mapping_required; failures return an exact reason (token expired, missing scope, site not selected, and so on). A Search Console site URL is not the same as a GA4 property id.
Related: Google Analytics 4 (GA4), Project Intelligence Datapoint, Bing
Google Analytics 4 (GA4)
Section titled “Google Analytics 4 (GA4)”Google’s analytics source — page views, sessions and engagement.
GA4 is a Google data source ingested once connected and a property is mapped to the project. It contributes pages and views/sessions/engagement over a date range as source-labelled datapoints. A GA4 property id must not be confused with a Search Console site URL.
Related: Google Search Console, Project Intelligence Datapoint
A user-owned search source — Bing queries, pages, clicks, impressions and CTR.
Bing is a self-serve search data source. If not configured it shows setup required; if configured but unmapped, mapping required; if mapped, it ingests available Bing search data or returns an exact no-data state. Repeated rows for the same query or URL are aggregated honestly by entity, and duplicate external keys never abort a batch insert.
Related: Google Search Console, Project Intelligence Datapoint
Keyword & SERP enrichment
Section titled “Keyword & SERP enrichment”Search volume, difficulty, SERP and competitor data added to keywords — only ever run on an explicit, credit-metered action.
Enrichment adds search volume, keyword difficulty, SERP layout and competitor intelligence to a keyword set. It is cost-sensitive, so it never runs on render or automatically during project creation — only when the user explicitly triggers it, metered in credits. The underlying data provider is managed by WordPresto at the platform level, not configured per user, and can be swapped without changing the workflow. Results are persisted as source-labelled datapoints; nothing is fabricated.
Related: Keyword Set, Project Intelligence Datapoint
Crawler
Section titled “Crawler”Core infrastructure that discovers what a website actually contains.
The crawler tells WordPresto what a site really contains: URLs, status codes, titles, meta descriptions, headings, canonical URLs, indexability, internal links, schema presence, page-type guesses, content sections and site structure. Its findings populate Project Intelligence and Site Health. When it cannot run automatically it exposes a clear action-required state rather than faking results.
Related: Site Health, Project Intelligence Datapoint
States
Section titled “States”Approval
Section titled “Approval”The gate content must pass before anything leaves the Canvas — persisted and auditable.
Approval means the required review Workers have run, relevant suggestions have been accepted, rejected or resolved, final editorial approval has been given and destination requirements have been checked. It is persisted and auditable (at minimum a status, an approvedAt timestamp and who approved), not client-only. Publishing agents must not write to external systems before approval, and editing the Canvas after approval returns it to draft/review and clears the persisted approval.
Related: Canvas, Draft, Publishing Agent
Content that has not been approved and must not be published.
Draft is the pre-approval state of Canvas content. AI-generated content is always marked draft until a human approves it, and editing approved content returns it to draft/review. Nothing in a draft state may be handed to a Publishing Agent.