Content Generation (Phase 3)

Goal: Generate well-written, SEO-optimized content in one shot.

Model Selection

Model	Best For	Notes
deepseek-v4-flash	Standard blog posts, supporting content	Fast, cost-effective
deepseek-v4-pro	Comprehensive guides, pillar pages, flagship content	Deep reasoning

Generation is done through opencode's built-in capabilities. The content prompt is passed directly to opencode, which handles the model interaction.

Generation Modes

Mode	When	Method
Single prompt	≤ 3,000 words	One structured prompt with keyword + NeuronWriter terms + guidelines
Section-by-section	> 3,000 words	Generate introduction first, then build each section with context carry-over

Output Format

Markdown file saved to:

content-md/{site}/{topic-category}/{PID}.md

Example:

content-md/fundamentalstock.com/stock-market-investing-financial-statements/FS-0001.md

Quality Control

warning

AI-generated content must be human-edited before publishing. Review for:

• Accuracy and factual correctness
• Natural term integration (not keyword stuffing)
• Original perspective and information gain
• Proper entity citations and source attribution
• E-E-A-T signals (experience, expertise, authority, trustworthiness)

After Generation → Publishing

Once content is generated and saved as markdown, the next step is automated publishing via wp-cli:

1. Markdown file sits at: content-md/{site}/{topic-category}/{PID}.md
2. A wp-cli script reads the markdown, converts to HTML, and creates a WordPress draft
3. Log into WordPress to add images, admonitions, internal links, and final formatting
4. Publish live when ready