Marketing knowledge packs: content, SEO and websites

What it covers

• Atomisation map - the core artifact: the atomisation map record shape (source, ordered pieces, per-piece fields: channel, format, core_claim, angle, internal_link_target, cta, sequence, lag_days). The standalone rule and the distinct-angles-not-paraphrases rule. One-source-many-pieces, and the no-padding rule (if a source has one real point, produce one derivative).
• Channel formats - generic channel-to-format taxonomy (canonical long-form, short social feed post, email teaser, etc.) and how each derivative differs from the source. One claim per piece.
• Linking and ctas - how to bake exactly one internal-link target and one CTA into each derivative piece. Target and anchor selection delegated to ../internal linking/_index. Hub-and-spoke: the long-form is always the hub.
• Sequencing - publish order and cadence: canonical piece first (sequence 1), derivatives lag; the no-orphan rule (every derivative links back to the hub); no two pieces share a verbatim core_claim; channel pacing.

What Corey can do with it

• Content atomisation - On-demand content-atomisation planner: reads a long-form source, builds a TSV atomisation map (sequence, channel, format, core_claim, internal_link_target, cta), validates it, computes a publish schedule, and emits a sequenced derivative plan with one internal-link target and one CTA per piece.
• Content performance - On-demand content-performance analyser: collect per-piece signals from up to four independent sources (analytics, GSC, crawler, LLM), run the deterministic scale/hold/cut scoring model, and emit a prioritised action list.
• Content pipeline - Orchestrate a multi-channel content brief from one idea.
• Weekly content loop - Monday orchestrator. Auto-picks the top-ranked fresh content idea from the content Calendar, drafts the full bundle (blog + 2-3 LinkedIn cuts + newsletter issue), writes one brief, packs it for approval via approval-pack, advances the idea to "Draft - Needs Approval", and stops.

Try asking

• Atomise this content
• Content performance
• Content pipeline

What it will not do

• No padding. If a source has one real point, produce one derivative and say so.
• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Performance signals - the four signal families in full: what each measures, how each is normalised to a +1/0/-1 vote, and neutral-vertical worked examples. Cross-links ../geo aeo/measurement for AI-citation measurement methodology.
• Decision rules - the deterministic decision model: per-signal vote logic, net-score-to-decision mapping, minimum-signal rule, threshold conventions, and the correlation-not-causation caveat.
• Data source seam - the clean seam: normalised per-piece input contract, optional independent sources, graceful degradation, missing-source reporting, and how a new source slots in. Cross-links ../keyword research/data source seam.

What Corey can do with it

• Content atomisation - On-demand content-atomisation planner: reads a long-form source, builds a TSV atomisation map (sequence, channel, format, core_claim, internal_link_target, cta), validates it, computes a publish schedule, and emits a sequenced derivative plan with one internal-link target and one CTA per piece.
• Content performance - On-demand content-performance analyser: collect per-piece signals from up to four independent sources (analytics, GSC, crawler, LLM), run the deterministic scale/hold/cut scoring model, and emit a prioritised action list.
• Content pipeline - Orchestrate a multi-channel content brief from one idea.
• Weekly content loop - Monday orchestrator. Auto-picks the top-ranked fresh content idea from the content Calendar, drafts the full bundle (blog + 2-3 LinkedIn cuts + newsletter issue), writes one brief, packs it for approval via approval-pack, advances the idea to "Draft - Needs Approval", and stops.

Try asking

• Atomise this content
• Content performance
• Content pipeline

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Scheduling and idempotency - deterministic scheduling doctrine: one-piece-per-slot cadence, hub-first ordering, the publish tuple definition, stable idempotency-key derivation, and what makes the queue deterministic.
• Channel seam - the pluggable-publisher and per-channel graceful-degradation seam: optional independent channels, surfaced (not silent) missing channels, and how a new channel or publisher backend slots in. Cross-links ../keyword research/data source seam and ../content performance/data source seam for the broader engine-wide seam pattern.

What Corey can do with it

• Multi channel publish - Deterministic, idempotent, dry-run-by-default publish scheduler.

Try asking

• Publish content schedule
• Talk me through scheduling and idempotency for Multi-channel publishing, and flag anything that changes often.
• What do I need to get right about channel seam before I start?

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Social channel operating - what each social surface type is for; personal vs brand account split; format tiers; posting and re-share cadence; the engagement-harvest loop; what not to do.
• Newsletter operating model - operating an owned newsletter: owned-vs-rented trade-off; destination-selection decision framework (three archetypes, recommendation heuristic); one-body-multiple-destinations pattern; send cadence and subject-line discipline.

What Corey can do with it

• Social visuals - Make a social image that is funny and informative rather than a grid of boxes.
• Newsletter - Draft a newsletter issue (hook-to-blog) from a canonical blog draft.

Try asking

• Make a card
• Newsletter
• Talk me through social channel operating for Social-channel and newsletter operating playbook, and flag anything that changes often.

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Volume and demand - estimating search demand without a paid tool: what each free source gives, its limits, and how to triangulate a demand estimate from all four.
• Search intent - the intent taxonomy (informational / commercial / transactional / navigational), deterministic signal tokens for each class, and why intent gates which content type to build.
• Clustering and gaps - grouping a keyword universe into topic clusters (shared-head method and SERP-overlap method), the gap-analysis method (demand minus own-coverage), and how one cluster maps to one target page to prevent keyword cannibalisation.
• Data source seam - the seam doctrine: the normalised keyword-signal record shape, graceful degradation when a source is unavailable, and how a paid provider slots in behind the same interface with zero downstream change.

What Corey can do with it

• Keyword research - On-demand keyword research that turns seed terms into a keyword map (expand -> volume -> intent -> cluster -> gap) on a free tool stack (GSC + Keyword Planner + pytrends + Firecrawl SERP).

Try asking

• Keyword research
• Talk me through volume and demand for Keyword research, and flag anything that changes often.
• What do I need to get right about search intent before I start?

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Answer engine surfacing - the GEO layer: how AI answer engines (conversational assistants, AI Overviews, Perplexity-class engines, agentic browsers) retrieve, ground, and cite sources - and what a site must do to be in the citation set.
• Answer shaped content - the AEO layer: extractable content patterns that make individual chunks liftable by a model - direct-answer-first structure, question-led headings, self-contained sections, explicit definitions, semantic HTML, and schema markup as an extractability aid.
• Citation worthiness - source-level signals that make content quotable over a competitor's: original data, demonstrable authority, specificity, freshness, and alignment with cross-source consensus.
• Measurement - tracking GEO and AEO performance: share-of-voice metrics, citation monitoring, brand-mention tracking, and how to instrument the free-stack alternatives to paid GEO-analytics platforms.
• Technical discoverability - the layer beneath the content: whether a page *can* be cited at all. Auditing the deployed site rather than the repo, the two-layer defect where a tag is missing from both schema and page component, build gates that stop it regressing, correct AI crawler agent names (there is no ChatGPTBot), sitemap/noindex agreement, markdown twins, and why the landing slug is the only attribution identifier that survives referer stripping.

What Corey can do with it

• Monthly geo review - Monthly audit of AI search visibility.

Try asking

• Monthly GEO review
• Talk me through answer engine surfacing for GEO + AEO, and flag anything that changes often.
• What do I need to get right about answer shaped content before I start?

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Internal link graph - building an internal-link graph from an own-site crawl: inbound vs outbound internal links, the adjacency model, orphan pages (zero inbound internal links), click depth from the homepage, and why an own-site crawl suffices for internal-link work.
• Anchor text - descriptive, varied, relevant anchors; over-optimisation and exact-match stuffing risks; anchor relevance to the target page's topic.
• Link density and clusters - internal-link density discipline (links per page relative to length), distribution, and the hub-and-spoke / pillar-and-cluster (topic cluster) model.

What Corey can do with it

• Internal linking - On-demand internal-linking audit that crawls the own site and produces a link-map, orphan-page list, per-page density table, and anchor-text suggestions.

Try asking

• Internal linking
• Talk me through internal link graph for Internal linking, and flag anything that changes often.
• What do I need to get right about anchor text before I start?

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Overview and islands - what Astro is (content-driven, server-first, zero-JS by default, MPA not SPA), the islands mental model, and when to choose it.
• Project structure and cli - the src/ layout, public/, astro.config.*, and the dev/build/preview CLI and npm scripts.
• Components and syntax - the .astro component (server script fence vs template), Astro.props + interface Props, expressions, and slots.
• Layouts - shared page shells, the Markdown layout prop, nested layouts via slot forwarding.
• Routing - file-based and dynamic routes, getStaticPaths(), rest params, pagination, and on-demand (SSR) routes.
• Endpoints and static files - custom .js/.ts endpoints and generating non-HTML files (the mechanism behind RSS, OG images and robots.txt).
• Content collections - the Astro 5 Content Layer: glob/file loaders, Zod schemas, image()/reference(), querying, rendering, generating pages, Markdown + MDX. The backbone of a markdown site.
• Islands and frameworks - the full client:* directive set and when to use each, the JS cost model, passing props/children, sharing state with nanostores, framework integrations, and server islands.
• And 17 more topics.

What Corey can do with it

• Firecrawl website design clone - Extract any website's design system into an agent-ready DESIGN.md using Firecrawl scrape evidence.
• Astro website builder - Take a business from nothing to a live Astro site - static or dynamic - scaffold, design, content, optionally wire dynamic features (database, auth, Stripe checkout, backend), build, test, deploy, and attach a custom domain.

Try asking

• Talk me through project structure and cli for Astro website builder, and flag anything that changes often.
• What do I need to get right about components and syntax before I start?

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Architecture and layout - the four-layer model and on-disk directory structure.
• Requirements and environment - PHP, database, HTTPS, and server baselines (2026).
• Request lifecycle - how a request bootstraps from index.php to template render.
• Hooks actions filters - the Plugin API: the core extensibility mechanism.
• Content model and database - posts, custom post types, taxonomies, and the schema.
• Themes templates and the loop - template hierarchy, the Loop, classic vs block themes.
• Block editor and fse - Gutenberg blocks and Full Site Editing.
• Rest API - JSON endpoints and authentication.
• And 3 more topics.

What Corey can do with it

Corey reads it whenever a job touches WordPress core, whichever skill is running.

Try asking

• Talk me through architecture and layout for WordPress, and flag anything that changes often.
• What do I need to get right about requirements and environment before I start?

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Wordpress platform and gpl - market position, the GPL licence that creates the plugin economy, and the three distribution channels.
• Hooks actions filters - the core extensibility primitive: actions vs filters, signatures, priority, accepted_args.
• Plugin anatomy lifecycle - plugin folder/header structure and the activation / deactivation / uninstall lifecycle contracts.
• Plugin security model - the shared-process trust model and the sanitise / escape / validate / nonce + capability baseline.
• Block editor blocks - the block (Gutenberg) editor as the modern UI-extension surface; block.json, static vs dynamic blocks.
• Woocommerce architecture hpos - WooCommerce as a plugin-on-a-platform: legacy custom post types, the CRUD/data-store abstraction, and High-Performance Order Storage.
• Woocommerce extensibility - extending WooCommerce in practice: hooks, CRUD queries, template overrides, REST vs Store API, payment gateways, declaring HPOS compatibility.
• Ecosystem risk supply chain - the vulnerability and supply-chain risk surface, and what it means for choosing and trusting plugins.

What Corey can do with it

Corey reads it whenever a job touches WooCommerce and the WordPress plugin model, whichever skill is running.

Try asking

• Talk me through wordpress platform and gpl for WooCommerce and the WordPress plugin model, and flag anything that changes often.
• What do I need to get right about hooks actions filters before I start?

What it will not do

• Quote a price, limit or threshold from memory. Figures that move are kept in a dated ledger and re-checked against the primary source first.
• Do anything public, costly or irreversible without your yes.

What it covers

• Adoption models
• Component catalog
• Composition patterns
• Licensing and fallback
• Reskin recipe

What Corey can do with it

• Catalyst portal builder - Build a product's authenticated platform/portal UI - the owner/admin dashboard, customer portal, or SaaS app shell - on the Catalyst UI kit (Tailwind Plus), reskinned to the business's design tokens.

Try asking

• Talk me through adoption models for Catalyst portal builder, and flag anything that changes often.
• What do I need to get right about component catalog before I start?

What it will not do

• Do anything public, costly or irreversible without your yes.