lp-full-generation

Generates complete 12-section landing pages using local Ollama LLMs with RAG and quality scoring.

Updated Mar 23, 2026
One-click install
npx skills add https://github.com/sakamotomomotaro0809-netizen/tateyomi --skill lp-full-generation-sakamotomomotaro0809-netizen
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: lp-full-generation
Source: https://github.com/sakamotomomotaro0809-netizen/tateyomi/tree/main/taisun_agent/.claude/skills/lp-full-generation
Command: npx skills add https://github.com/sakamotomomotaro0809-netizen/tateyomi --skill lp-full-generation-sakamotomomotaro0809-netizen

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Writing a complete, high-converting landing page section by section is slow and inconsistent. This Skill automates the full pipeline: it generates all 12 sections of a Taiyo-style LP using local LLMs, scores each section for quality, regenerates weak sections, and saves the result—entirely offline at zero API cost. ## Core Features & Use Cases - Full 12-Section Pipeline: Generates headline, lead, problem, agitation, solution, benefit, bullet, proof, story, offer, CTA, and P.S. sections sequentially using RAG retrieval from ChromaDB and Ollama models (qwen3:8b, qwen2.5:32b). - Automatic Quality Assurance: Each section is scored with taiyo-analyzer; sections below 70 points are automatically regenerated (up to 2 retries), targeting an overall score of 80+. - Knowledge Feedback Loop: Finished LPs are saved to Obsidian and added back to ChromaDB to enrich future generation. - Use Case: A marketer needs a complete LP for an AI agent training school targeting aspiring side-business professionals. One command produces the full 12-section draft with per-section scores and a generation report in about 10-20 minutes. ## Quick Start Ask the AI to generate a full LP for your product by specifying the target audience, product name, and tone, for example: generate a full LP for an AI agent school targeting business professionals interested in side businesses.

Frequently Asked Questions about lp-full-generation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a full landing page with a local LLM?▼

Run the generate-with-quality.py script with the full argument, passing --target, --product, and --tone parameters. It generates all 12 sections sequentially using Ollama models with RAG retrieval, then scores and saves the result automatically.

What Ollama models are needed for LP generation?▼

The pipeline uses qwen3:8b for shorter sections like headline, bullet, CTA, and P.S., and qwen2.5:32b for longer persuasive sections like lead, problem, and story. The 32b model requires roughly 20GB of VRAM.

How does the automatic quality scoring and regeneration work?▼

Each generated section is scored immediately with taiyo-analyzer. Sections scoring below 70 points are regenerated with improvement feedback, up to 2 retries. The full LP targets an overall score of 80 or higher before being saved.

Does this LP generation require an internet connection or API keys?▼

No, the entire pipeline runs locally at zero cost. It uses Ollama for model inference and a local ChromaDB instance for RAG retrieval, so no external API keys or network access are required.

How long does full 12-section LP generation take?▼

On an M1/M2 Mac, the four qwen3:8b sections take about 2-4 minutes and the eight qwen2.5:32b sections take about 8-16 minutes, totaling roughly 10-20 minutes. Regeneration of low-scoring sections adds extra time.

What are the prerequisites before running the LP generation pipeline?▼

Ollama must be running with the required models pulled, ChromaDB must be indexed with LP knowledge chunks, and the Python virtual environment must be activated. Insufficient memory for the 32b model is a common failure point.