LandingForge

AI-powered B2B landing page generator using a local LLM via @llama-node/core.

Setup

1. Install dependencies

npm install

2. Add a compatible local model

This project uses @llama-node/core and expects a GGML model (typically .bin).

Auto-discovery checks these paths first:

models/model.bin
models/model.ggml
models/model.GGML

If multiple model files exist, set an explicit path:

# PowerShell
$env:LLM_MODEL_PATH = "models/model.bin"

Important:

• GGUF is not supported by this codebase/runtime.
• Some files named .GGML are actually GGUF internally; if the app reports Unsupported model format: GGUF, replace it with a true GGML .bin model.

3. Run the dev server

npm run dev

Open http://localhost:3000.

Usage

1. Fill in Company Name and Short Description (required).
2. Optionally add Website URL and/or paste content into Website Text / Research Notes.
3. Choose a Tone.
4. Click Generate Landing Page.
5. View the live Preview, inspect the Content JSON, or copy/download the generated HTML.

What Changed (Design + Research)

• Each generation now uses an auto-selected visual variant (different font pair, palette, motion profile, and section treatment).
• Recent visual variants are avoided to reduce back-to-back repetition.
• If a website_url is provided, the API now fetches and summarizes key public HTML text (title, meta description, headings, paragraphs, list items) and merges it with pasted notes before prompting the model.

Project Structure

landingforge/
  app/
    api/generate/route.ts   <- API route: calls local LLM + website research fetch
    layout.tsx
    page.tsx
  components/
    LandingPageAgent.tsx    <- Main UI component
  lib/
    llm.ts                  <- Local LLM wrapper
    prompt.ts               <- System prompt + helpers
    research.ts             <- Website text preprocessing + URL fetch summarizer
    showcase-template.ts    <- Multi-variant landing page HTML renderer
  models/
    model.bin               <- Place your local model here (not in git)
  next.config.mjs
  package.json
  tsconfig.json

Tech Stack

• Next.js 14 (App Router)
• @llama-node/core
• TypeScript
• React 18