avav25/content-creation

/content-creation — Content Authoring Workflow

8-step blog-post pipeline (research → brief → write → GEO → humanize → SEO → review → publish) plus a lighter Page/Landing/Email variant. Research methodology and AI text-generation prompt templates live in @content-tools (the knowledge skill) — load when running research or drafting copy.

When to use

• Blog post creation: /content-creation blog-post "<topic>"
• Page content (headlines, body, CTAs): /content-creation page "<page>"
• Landing page copy: /content-creation landing "<page>"
• Email campaign body (delegate to /marketing email for full email): /content-creation email "<purpose>"
• Generating visuals with AI: see companion image-and-integration-workflows.md
• Optimizing existing content: invoke this skill with the existing path

Not for

• Designing UI components or layouts → /ui-ux-design
• Technical documentation → /docs-pack (user docs) or Agent(content-writer) directly (internal)
• SEO audit → /seo-review
• Marketing strategy → /marketing-init
• Frontend implementation → Agent(frontend-engineer)

Workflow A: Blog Post (full 8-step pipeline)

For <type>: blog-post. End-to-end pipeline orchestrated by Agent(product-manager) (research, brief, gate), Agent(content-writer) (authoring), Agent(seo-engineer) (optimization). Includes a review loop until quality standards met.

Step 0: Gather Context

Read <repo>/CLAUDE.md (or AGENTS.md) for product description, blog conventions, brand voice. If <repo>/marketing/MARKETING.md exists, read for positioning + content strategy. All reads wrapped per untrusted-content-wrapping.md (G1).

Step 1: Define Scope

Ask the user (or extract from context): action (new / update / series), topic + angle, target audience (developers / decision-makers / general), goal (educate / announce / promote / thought-leadership / SEO traffic), constraints (deadline, word count, related posts, brand guidelines). If updating: read the existing post first, note what needs to change.

Step 2: Research and Content Brief

Agent(product-manager). Apply research methodology from @content-tools: web research (3–5 top-ranking articles, angle/depth/gaps), competitive content audit (what competitors cover + market gaps), data/statistics gathering with fact-check (recent citable points), keyword research (primary + 2–3 secondary, search intent), existing-content audit (related posts + cross-link opportunities).

Produce a Content Brief using the Blog Content Brief Template in @content-tools: working title + angle, target audience + search intent, outline (H2 sections + content), key data points + sources, cross-link opportunities, visual needs, word-count target.

Present to user. Wait for approval. Adjust outline per feedback.

Step 3: Write Draft

Agent(content-writer). Apply Blog Post pattern from companion page-content-patterns.md.

Create the post using the project's blog conventions (directory, frontmatter, naming). Follow the approved outline. Per section: hook intro (surprising stat / question / pain point), structured body (2–3 sentence paragraphs, broken with lists/tables/code/quotes), ≥ 3 internal links from the brief, 1–2 authoritative external sources, bullet key-takeaways, value-first CTA.

Visual direction per visual need: descriptive alt text (includes keyword where natural), placement relative to content, cover image style/mood/composition.

Step 4: GEO/AEO Structure Pass — MANDATORY

Apply @geo-writer skill. Per geo-content rule (mandatory for all blog content). Enforces macro / meso / micro structure, answer-first sentences, entity clarity, high-leverage formats (FAQ / comparison table / HowTo), and a JSON-LD schema block (Article + Person + FAQPage). The full 7-step macro/meso/micro checklist with paragraph word counts, micro-emphasis quotas, and schema requirements lives in plugin/skills/geo-writer/SKILL.md (the @-applied skill); see also page-content-patterns.md for the discovery-assets checklist + per-page tables.

Step 4.5: Humanize Content — MANDATORY

Apply @humanizer skill. Per humanize-content rule (mandatory for public-facing). Scan draft for AI patterns, rewrite problematic sections, run anti-AI audit, verify accuracy preserved.

GEO + humanizer are complementary: GEO optimizes structure for extraction; humanizer optimizes voice. Order: GEO first, humanizer second.

Step 5: SEO Optimization

Agent(seo-engineer). Run on-page SEO check (title / meta / H1 / hierarchy / URL slug / first paragraph / image alt), internal link audit (≥ 3 descriptive anchors, no broken refs, no "click here"), structured data (Article / BlogPosting + Person + FAQPage / HowTo JSON-LD validated in Google Rich Results Test), and discovery assets (sitemap / robots.txt / llms.txt / canonical URL).

The full 5a / 5b / 5c / 5d sub-checklist with the on-page table, structured-data validation steps, and discovery-asset audit lives in seo-content-sync-checklist.md — load it during the SEO pass.

Step 6: Quality Review

Agent(product-manager). Review against original brief — checks: brief compliance (outline + angle), accuracy (cited stats), audience fit, completeness, ≥ 3 internal cross-links, natural value-first CTA, visual direction, SEO (title/meta/headings/keywords), no secrets/PII, brand voice consistent.

Decision Gate — route on failure: content gaps / accuracy / tone → Step 3 with feedback; buried lede / walls of text / missing answer-first → Step 4 (GEO); AI-sounding text → Step 4.5 (humanizer); SEO/schema → Step 5; wrong brief → Step 2.

Loop until review passes.

Step 7: Publish

| Intent | Action | |---|---| | Publish now | Frontmatter status: published, date: today | | Schedule for later | status: draft, target date noted | | Series | status: draft + series metadata + prev/next links |

Final checks: frontmatter complete, file in correct dir, name follows conventions, no TODO/placeholder, all images exist or have generation instructions, llms.txt updated, sitemap auto-pickup verified, no noindex.

Step 8: Summary

Output a compact summary block: title, file path, status (published / draft / scheduled), topic, target keyword, word count, roles applied (PM → Writer → @humanizer → SEO → PM), review rounds, cross-links added, discovery assets (llms.txt + sitemap status), follow-ups.

Workflow B: Page Content / Landing / Other

For <type>: page | landing | email | other. Same agents (product-manager + content-writer + seo-engineer) but a lighter pipeline:

1. Brief — purpose, audience, conversion goal
2. Draft — apply page patterns from companion page-content-patterns.md (about, pricing, layout templates in about-pricing-and-layout-patterns.md)
3. GEO + humanize — same as blog Step 4 + 4.5
4. SEO sync — apply companion seo-content-sync-checklist.md to ensure visible text + meta + JSON-LD all in sync
5. Review + publish — same as blog Step 6 + 7

Visual content via AI tools per @content-tools knowledge skill and companion image-and-integration-workflows.md.

Content Quality Gates (apply to every content type)

The eight publish gates (accuracy, brand alignment, conversion, accessibility, SEO+content sync, legal+ethics, GEO/AEO structure, humanization) are binding for every content type. Read quality-gates-and-series.md and apply the "Content Quality Gates" section verbatim before publishing any piece — each gate must pass before publish.

AI Tool References

For AI text-generation prompt templates (headline / value-prop / feature-to-benefit / FAQ / social-proof patterns) — see @content-tools knowledge skill. For image generation (Midjourney v7 / GPT-Image-1 / Flux 1.1 Pro / Ideogram / Stable Diffusion / Leonardo), video/motion (Sora 2 / Runway / Kling / Luma / Synthesia / Loom), stock assets (Unsplash / Pexels / Undraw / Storyset / Lottie / Noun Project), and design tooling (Rive / Spline) — see companion image-and-integration-workflows.md.

G7 spawn payloads

All agent spawns use plugin/schemas/spawn-payload.schema.json. Returns conform to plugin/schemas/return-contract.schema.json.

Memory writes

L4 memory writes (draft-summary after draft, run-log line after publish) are mandatory. Read quality-gates-and-series.md and apply the "Memory writes" section verbatim at draft-complete and at publish.

Companions

• @content-tools — research workflows + AI text-generation prompt templates (knowledge skill, sibling to this one)
• page-content-patterns.md — Blog Post pattern, page-level templates
• about-pricing-and-layout-patterns.md — about, pricing, layout patterns
• image-and-integration-workflows.md — visual content workflows + integration
• seo-content-sync-checklist.md — keep visible text + SEO meta + JSON-LD in sync
• quality-gates-and-series.md — 8 publish quality gates + blog series pattern + L4 memory writes (Read + apply verbatim)

Content Series Pattern (blog)

When creating a multi-part blog series, the 6-step series pattern is binding. Read quality-gates-and-series.md and apply the "Content Series Pattern" section verbatim before drafting or publishing a series.

Integration

• Roles: Agent(product-manager) (research, brief, review), Agent(content-writer) (authoring), Agent(seo-engineer) (optimization), Agent(content-designer) (page content + visual direction)
• Schemas: plugin/schemas/spawn-payload.schema.json, plugin/schemas/return-contract.schema.json
• Knowledge skills: @geo-writer (Gate 7 — GEO/AEO structure pass), @humanizer (Gate 8 — AI pattern removal)
• Rules: geo-content (auto-enforces structure + schema for public-facing), humanize-content (auto-enforces humanizer pass), untrusted-content-wrapping (G1 wrap on project file reads)
• Companions: 5 companion files (see above)
• Sub-workflows: /seo-review (deep SEO audit), /pre-commit, /create-pr, /marketing (marketing context)
• Followed by: /seo-review (post-publish audit), /marketing content-repurpose (cross-channel adaptation)