Myvar Website

A note-keeping website that transforms brief notes into polished articles using AI agents.

I use this project to create content about technologies and software engineering for my website.
https://gibbok.github.io/myvar/

What it does

Send draft notes to the system, and AI agents automatically generate, review, and publish website articles.

Tech Stack

• Architecture: JAMstack with Hugo hosted on GitHub Pages
• AI Orchestration: LangGraph with Google Gemini LLM
• Search: Pagefind - fully static search library
• Agents:
  • Generator: Creates content from draft notes
  • Reviewer: Reviews and provides feedback for improvements
  • Publisher: Finalizes content and generates metadata for publication

How it works

1. For local generation, write notes in generator/drafts/content.md
2. Run the generator: cd generator && make start
3. Agents generate, review, and publish the article
4. Output is written to website/content/
5. OG images for social sharing are created automatically at build time

Setup

# Install dependencies
uv sync

# Configure environment
cp .env.example .env
# Add your GEMINI_API_KEY to .env

Usage

# Generate content locally
cd generator && make start

The generator/drafts/ folder is only for local content creation and should not be committed.

You can also generate content in GitHub:

1. Open Actions.
2. Run Generate content.
3. Paste your notes into the content input.
4. The workflow generates the article and opens a draft PR with changes in website/content/.

Merge the PR to deploy the site to GitHub Pages.

Makefile Commands

Generator (generator/)

• make start - Run the AI agent pipeline to generate articles

Website (website/)

• make serve - Start Hugo development server
• make build - Build the static site

Environment Variables

• GEMINI_API_KEY: Your Google Gemini API key
• GEMINI_MODEL: Model to use (default: gemini-2.5-flash)
• DEBUG_MODE: Enable debug output (default: false)