NeuronWriter → WordPress Content Pipeline

Status: Draft for review
Last updated: 2026-07-28
Repo: rahul-10xers/ClaudeCode2.0 (source of truth for all scripts & guides)

The Big Picture — Mermaid Flow
What You Do vs What I Do
The 4 Page Types
How ACF Field Injection Works
NW API vs NW MCP — What Each Unlocks
The Complete Pipeline — Step by Step
Credentials & Auth
Open Questions for You

1. The Big Picture — Mermaid Flow

flowchart TB
    subgraph YOU["🔵 You (Human)"]
        A[Pick a keyword & page type] --> B[Create NW query<br/>app.neuronwriter.com<br/>1 analysis credit]
        B --> C[Share query URL/ID]
        C --> D[Upload images to WP<br/>Media Library, share IDs]
    end

    subgraph MCP["🟢 MCP Layer (free API calls)"]
        E[list-projects] --> F[new-query<br/>if not already created]
        F --> G[get-recommendations<br/>poll until ready]
        G --> H[get-competitors-content<br/>read competitor H2/H3]
    end

    subgraph BROWSER["🟡 Browser Layer (one session per article)"]
        I[Open NW query URL] --> J[Step 2: Select competitors]
        J --> K[Step 3: Generate meta title/desc]
        K --> L[Step 4: Open Content Designer Wizard]
        L --> M[Step 5: Paste content brief]
        M --> N[Step 6: Review outline]
        N --> O[Step 7: Extract facts from URLs]
        O --> P[Step 8: Generate article<br/>wait ~120s]
        P --> Q{Score ≥ 70?}
        Q -->|No| R[Step 9: Optimize<br/>Smart Headings → Auto-insert → FAQ]
        R --> Q
        Q -->|Yes| S[Step 10: Plagiarism check<br/>optional]
    end

    subgraph API["🟣 API Layer (direct REST calls)"]
        T[get-editor-content<br/>via MCP or NW API] --> U[Build WP payload<br/>by page type]
        U --> V[POST to WP REST API]
        V --> W[Re-save draft<br/>ACF fields need this]
        W --> X[Update Google Sheet<br/>tracker]
    end

    A --> MCP
    C --> BROWSER
    D --> API
    MCP --> BROWSER
    BROWSER --> API
    X --> Y[✅ Done<br/>Draft ready for review]

2. What You Do vs What I Do

You Do (human-only, cannot be automated)

Step	Why you must do it
Pick a keyword + page type	Strategic decision — comparison, feature, free tool, or blog article?
Create the NW query	Costs 1 of 75 monthly analysis credits. I cannot spend your credits without your approval.
Upload images to WP Media Library	Images need a human to select/upload. I cannot invent image IDs.
Review the final draft	The pipeline produces a draft — you review and publish.
Decide on new page types	If we need a new template, you define the ACF fields.

I Do (fully automated)

Phase	What happens
MCP calls	List projects, create query (if you approve), poll until ready, read recommendations & competitor content
Browser automation	Drive the NW Content Designer wizard — select competitors, generate meta, paste brief, generate article, optimize score to ≥70
API calls	Read final content from NW, build the correct payload for the page type, POST to WordPress, re-save draft, update Google Sheet
Payload assembly	Map NW’s HTML output to the correct ACF fields for each page type (see section 4)

Hybrid (I do, but you provide input)

Step	What I need from you
Content brief tokens	Feature name, what it does, target persona (for feature pages)
Competitor name	For comparison pages
Article topic + keywords	For blog articles
WP page ID	If updating an existing page (from sheet col U)

3. The 4 Page Types

	Comparison	Free Tool	Feature	Article (Blog)
Template	`page-comparison.php`	`page-free-tools.php`	`page-features.php`	`single.php`
WP type	page	page	page	post
ACF prefix	`comp_*`	`ft_*`	`feat_*`	Theme meta fields
Category	18	46	9	3
NW content type	`landing-page`	`landing-page`	`landing-page`	`blog`
Competitor type	landing pages	landing pages	landing pages	blog/informational
WP endpoint	`wp/v2/pages`	`wp/v2/pages`	`wp/v2/pages`	`gr-articles/v1/create`
Body content	ACF fields only	ACF fields only	ACF fields only	`article_sections[]` → block HTML
Publisher script	✅ `nw-to-comparison-page.js`	✅ `nw-to-free-tool-page.js`	❌ Manual (script TBD)	❌ Manual (script TBD)
Content source	NW article → curated into ACF	NW article → auto-parsed by `nw-html-to-acf.js`	NW article → manually curated into `feat_*` fields	NW article → mapped to `article_sections[]` patterns

4. How ACF Field Injection Works

Comparison Pages (`comp_*` fields)

The script nw-html-to-comparison-acf.js parses NW’s HTML output:

NW HTML (H1, H2, H3, P, LI, TABLE tags)
    │
    ▼
Regex tokenizer strips tables, lists, empty paragraphs
    │
    ▼
Maps sections by H2 text patterns:
  - "GrowReach vs [X]" → comp_comparison_table rows
  - "Key Features" → comp_guide_sections[]
  - "FAQ" → comp_faq_items[]
  - Any other H2 → comp_guide_sections[]
    │
    ▼
Merged with competitor-file.js data (pricing, features, positioning)
    │
    ▼
Full payload: { title, status, template, page_category, acf: { comp_* } }
    │
    ▼
POST /wp/v2/pages → re-save draft

Key ACF fields:

comp_hero_title, comp_hero_desc, comp_hero_badge
comp_comparison_table (rows of feature-vs-competitor)
comp_guide_sections[] (heading + content pairs)
comp_faq_items[] (question + answer)
comp_pricing_cards[] (plan name, price, features)
comp_cta_text, comp_cta_link

Free Tool Pages (`ft_*` fields)

The script nw-html-to-acf.js parses NW’s HTML output:

NW HTML (H1, H2, H3, P, LI tags)
    │
    ▼
Regex tokenizer splits by H2 markers
    │
    ▼
Routes by H2 text pattern (CRITICAL — exact phrasing required):
  - "How [Tool] Works" → ft_features_heading + ft_features_cards[]
    (H3s with number emojis 1️⃣2️⃣3️⃣ become cards with icons)
  - "Frequently Asked Questions" → ft_faq_heading + ft_faq_items[]
  - Any other H2 → ft_guide_sections[]
    │
    ▼
Full payload: { title, status, template, page_category, acf: { ft_* } }
    │
    ▼
POST /wp/v2/pages → re-save draft

Key ACF fields:

ft_hero_title, ft_hero_desc, ft_hero_badge, ft_hero_stats[]
ft_features_heading, ft_features_cards[] (icon, title, desc)
ft_guide_sections[] (heading + content)
ft_faq_heading, ft_faq_items[]
ft_tool_html (the embedded tool iframe)
ft_cta_text, ft_cta_link

⚠️ Critical rule: The H2 phrasing “How [Tool] Works” and “Frequently Asked Questions” must be exact — the regex parser depends on these strings. If NW paraphrases, content lands in the wrong ACF field.

Feature Pages (`feat_*` fields)

No auto-parser exists yet — content is manually curated from NW’s article:

NW article HTML
    │
    ▼
Manual extraction by H2 section labels:
  "HERO SECTION" → feat_hero_title, feat_hero_desc, feat_hero_benefits[]
  "CAPABILITY 1:" → feat_capabilities_list[0] (heading + desc + layout)
  "CAPABILITY 2:" → feat_capabilities_list[1]
  "CAPABILITY 3:" → feat_capabilities_list[2]
  "CAPABILITY 4:" → feat_capabilities_list[3]
  "COMPARISON SECTION:" → feat_comp_engage_* fields
  "RESULTS AND STATS SECTION" → feat_results_stats_items[]
  "USE CASE EXAMPLES SECTION" → feat_keyword_examples_items[]
  "TARGET ROLES SECTION" → feat_roles_items[]
  "FAQ SECTION" → feat_faq_items[]
    │
    ▼
Full payload: { title, slug, status, template, page_category,
    meta: { rank_math_* }, acf: { feat_* } }
    │
    ▼
POST /wp/v2/pages → re-save draft

Key ACF fields:

feat_hub_icon, feat_hub_desc, feat_hub_benefits — required (without these, Feature Hub shows “Coming Soon”)
feat_hero_title, feat_hero_desc, feat_hero_benefits[], feat_hero_btn_text/link, feat_hero_media_type, feat_hero_hero_image
feat_capabilities_list[] — 4 items, layouts alternate: left, right, left, right
feat_comp_engage_* — comparison table (neg vs pos columns)
feat_results_stats_items[] — 4 stat cards with emoji icons
feat_keyword_examples_items[] — 4 persona cards (themes: blue/purple/green/orange)
feat_roles_items[], feat_faq_items[]
feat_show_target: false, feat_show_clone_button: true

Also includes Rank Math SEO in meta object:

rank_math_title, rank_math_description, rank_math_focus_keyword, rank_math_canonical_url

Article Pages (Blog Posts) — `article_sections[]`

Uses a custom endpoint — NOT wp/v2/posts:

NW article HTML
    │
    ▼
Manual mapping to 9 content patterns:
  Each H2 section → one article_sections[] item
    │
    ▼
Full payload: { title, status, categories: [3], featured_media,
    content: "",  ← ALWAYS empty string
    meta: { hero_badge, hero_short_desc, post_highlights_*,
            faq_*, cta_* },
    article_sections: [ ... ] }
    │
    ▼
POST /wp-json/gr-articles/v1/create
    │
    ▼
PHP converter builds post_content from article_sections[]
    │
    ▼
Re-save draft via /wp/v2/posts/<id>

The 9 content patterns for article_sections[]:

#	Pattern	When to use
1	`TEXT ONLY`	H2 + one paragraph (intro)
2	`TEXT + H3 SUB-SECTIONS`	H2 with multiple sub-points
3	`TEXT + LIST`	Paragraph + bullet list
4	`TEXT + IMAGE`	Paragraph + full-width image
5	`TABLE ONLY`	Pure data comparison table
6	`TABLE + IMAGE`	Table + image
7	`TEXT + TABLE + IMAGE`	Full rich-media section
8	`STEPS`	Numbered step-by-step
9	`MULTIPLE IMAGES`	Before/after comparison

⚠️ Critical: content: "" must always be empty. The PHP class Article_Section_Blocks::build_rest_payload() converts article_sections[] into WordPress block HTML server-side. Using wp/v2/posts instead of gr-articles/v1/create produces a blank post body.

5. NW API vs NW MCP — What Each Unlocks

NW REST API (existing, via `neuronwriter-api.js`)

Endpoint	Cost	What it does
`POST /list-projects`	0	List your projects
`POST /new-query`	1 credit	Create a new SERP analysis
`POST /get-query`	0	Get analysis data (NLP terms, competitors, questions)
`POST /list-queries`	0	List queries with filters
`POST /get-content`	0	Get saved editor content (HTML + title + desc)
`POST /evaluate-content`	0	Score HTML without saving
`POST /import-content`	0	Save HTML to editor (creates revision)

Auth: X-API-KEY: n-cf40ec62b9297baad6e262cded11d808

NW MCP Server (new — `https://app.neuronwriter.com/mcp/http`)

Same endpoints as the REST API, but accessible via MCP protocol. The MCP adds:

Advantage	What it means
OAuth login	No API key management — login via browser once
`get-competitors-content`	Read competitor H2/H3 structures — not available in the old REST API
`source: "neuron-mcp"` filter	Queries created via MCP are tagged separately
Standard MCP interface	Any MCP client (Claude Desktop, Cursor, etc.) can use it

What MCP does NOT replace:

❌ Content Designer wizard (Steps 4-8) — still needs the browser
❌ Smart Headings / Auto-insert / FAQ optimizer (Step 9) — UI-only features
❌ The actual AI article generation — NW’s Content Designer does this

How I’ll Use Both

MCP (free reads)          REST API (writes)          Browser (generation)
    │                          │                          │
    ├─ list-projects           ├─ import-content          ├─ Content Designer wizard
    ├─ new-query (1 credit)    ├─ evaluate-content        ├─ Smart Headings
    ├─ get-recommendations     │                          ├─ Auto-insert terms
    ├─ get-competitors-content │                          ├─ FAQ generation
    ├─ get-editor-content      │                          └─ Score optimization
    └─ get-queries             │
                               │
                    Both hit the same backend:
                    https://app.neuronwriter.com/neuron-api/0.5/writer

6. The Complete Pipeline — Step by Step

Phase 0: Setup (one-time)

Register NW MCP server in Hermes SEO profile
Verify NW API key works (node -e "require('./scripts/lib/neuronwriter-api').listProjects().then(d=>console.log(d[0]?.name))")
Verify WP REST access (curl -w "%{http_code}" https://growreach.app/wp-json/wp/v2/pages?per_page=1 -H "Authorization: Basic ...")

Phase 1: You Start

Step	Who	What
1	You	Pick a keyword + decide page type (comparison/feature/free-tool/article)
2	You	Create the NW query at `app.neuronwriter.com` (costs 1 analysis credit)
3	You	Share the query URL/ID with me
4	You	Upload any needed images to WP Media Library, share the attachment IDs

Phase 2: MCP Reads (free, no browser)

Step	Who	What
5	Me	Call `get-recommendations` via MCP — poll until status = “ready”
6	Me	Read NLP terms, target word count, PAA questions, competitor scores
7	Me	Call `get-competitors-content` — read competitor H2/H3 structures

Phase 3: Browser — Content Designer Wizard (one session)

Step	Who	What
8	Me	Open NW query URL in browser
9	Me	Login check + dismiss driver.js tour
10	Me	Run state check to determine where to resume
11	Me	Select competitors (score > 50, 5-7 minimum)
12	Me	Generate meta title & description via NW’s AI
13	Me	Open Content Designer Wizard (NEURO chevron → Long-Form Content Designer)
14	Me	Paste content brief (varies by page type)
15	Me	Set tone = Expert, words = 2000, paste 3-persona template
16	Me	Review outline, inject TLDR H2 if missing
17	Me	Extract facts from top 5 competitor URLs (wait ~60s)
18	Me	Generate article (wait ~120s)
19	Me	Save and check score

Phase 4: Browser — Optimize to ≥70

Step	Who	What
20	Me	If score < 70: Smart Headings → add quality-checked H2s
21	Me	Auto-insert unused terms (up to 2 attempts)
22	Me	FAQ with Unused Terms (if still < 70)
23	Me	Re-check score. Target: beat best competitor by a few points.

Phase 5: API — Publish to WordPress

Step	Who	What
24	Me	Read final content via MCP `get-editor-content`
25	Me	Build WP payload matching the page type’s ACF structure
26	Me	POST to correct WP endpoint
27	Me	Re-save draft (mandatory — ACF fields don’t render without it)
28	Me	Update Google Sheet tracker (one write, cols F-X)

Phase 6: You Review

Step	Who	What
29	You	Preview the draft in WP admin
30	You	Make edits or approve for publish

7. Credentials & Auth

WordPress REST API

Endpoint: https://growreach.app/wp-json/
Auth:     Basic (base64 of "claudecode-rahul:AsiN dlhT hPC9 ZC4E k2qM yLVa")

NeuronWriter API

Endpoint: https://app.neuronwriter.com/neuron-api/0.5/writer
Auth:     X-API-KEY: n-cf40ec62b9297baad6e262cded11d808
Project:  growreach.app (ID: 77b8d4a7acd88c2d)
Plan:     Gold (WordPress + GSC + API access)
Credits:  75 analysis/month, 45,000 AI credits/month

NeuronWriter MCP

Server URL: https://app.neuronwriter.com/mcp/http
Auth:       OAuth (login via browser, approve connection)

Google Sheets (Tracker)

Spreadsheet ID: 13q9uw9up_w0oe8M4jyTQX0NBPp2zfqwu43yiIPWbXVw
Auth:           Service account JSON (in CREDENTIALS.md)

8. Open Questions for You

Should I register the NW MCP server in Hermes? This would let me call MCP tools directly instead of using the REST API client.
For feature pages — the publisher script (nw-to-feature-page.js) doesn’t exist yet. Should I build it? It would auto-parse NW’s H2-sectioned output into feat_* ACF fields, similar to how nw-html-to-acf.js works for free tool pages.
For article pages — same situation. Should I build nw-to-article-page.js that maps NW content to the 9 article_sections[] patterns?
Batch mode — do you want a mode where I take a list of keywords + page types and process them one by one without you needing to kick off each one?
Which profile should run this? The SEO profile doesn’t exist yet. Should I create one, or should the default profile handle it?
How do you want to trigger this? Via a kanban task? A cron job? A chat command like “write a feature page for [keyword]”?

Explorer

Neuronwriter Wordpress Pipeline Overview

NeuronWriter → WordPress Content Pipeline

Table of Contents

1. The Big Picture — Mermaid Flow

2. What You Do vs What I Do

You Do (human-only, cannot be automated)

I Do (fully automated)

Hybrid (I do, but you provide input)

3. The 4 Page Types

4. How ACF Field Injection Works

Comparison Pages (`comp_*` fields)

Free Tool Pages (`ft_*` fields)

Feature Pages (`feat_*` fields)

Article Pages (Blog Posts) — `article_sections[]`

5. NW API vs NW MCP — What Each Unlocks

NW REST API (existing, via `neuronwriter-api.js`)

NW MCP Server (new — `https://app.neuronwriter.com/mcp/http`)

How I’ll Use Both

6. The Complete Pipeline — Step by Step

Phase 0: Setup (one-time)

Phase 1: You Start

Phase 2: MCP Reads (free, no browser)

Phase 3: Browser — Content Designer Wizard (one session)

Phase 4: Browser — Optimize to ≥70

Phase 5: API — Publish to WordPress

Phase 6: You Review

7. Credentials & Auth

WordPress REST API

NeuronWriter API

NeuronWriter MCP

Google Sheets (Tracker)

8. Open Questions for You

Graph View

Table of Contents

Explorer

Neuronwriter Wordpress Pipeline Overview

NeuronWriter → WordPress Content Pipeline

Table of Contents

1. The Big Picture — Mermaid Flow

2. What You Do vs What I Do

You Do (human-only, cannot be automated)

I Do (fully automated)

Hybrid (I do, but you provide input)

3. The 4 Page Types

4. How ACF Field Injection Works

Comparison Pages (comp_* fields)

Free Tool Pages (ft_* fields)

Feature Pages (feat_* fields)

Article Pages (Blog Posts) — article_sections[]

5. NW API vs NW MCP — What Each Unlocks

NW REST API (existing, via neuronwriter-api.js)

NW MCP Server (new — https://app.neuronwriter.com/mcp/http)

How I’ll Use Both

6. The Complete Pipeline — Step by Step

Phase 0: Setup (one-time)

Phase 1: You Start

Phase 2: MCP Reads (free, no browser)

Phase 3: Browser — Content Designer Wizard (one session)

Phase 4: Browser — Optimize to ≥70

Phase 5: API — Publish to WordPress

Phase 6: You Review

7. Credentials & Auth

WordPress REST API

NeuronWriter API

NeuronWriter MCP

Google Sheets (Tracker)

8. Open Questions for You

Graph View

Table of Contents

Comparison Pages (`comp_*` fields)

Free Tool Pages (`ft_*` fields)

Feature Pages (`feat_*` fields)

Article Pages (Blog Posts) — `article_sections[]`

NW REST API (existing, via `neuronwriter-api.js`)

NW MCP Server (new — `https://app.neuronwriter.com/mcp/http`)