article_writer

LLM-powered article generator built with LangChain and OpenAI.

A modular, OOP-based pipeline that takes high-level constraints and produces a fully-structured article — table of contents → outlines → sections → title → optional self-critique pass.

Architecture

article_generator/
├── article/                  # Article model
├── paragraph/                # Paragraph + Resource models
└── article_generator/
    ├── article_generator.py  # Main orchestrator
    └── prompts.py            # LangChain PromptTemplate factories

Pipeline

1. Generate table of contents from a title
2. Generate outlines for each section (what to include / exclude)
3. Generate contents for each section
4. Generate the final title
5. Optional self-critique pass

Setup

python -m venv venv
source venv/bin/activate   # Windows: venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env       # then add your OPENAI_API_KEY

Usage

from article_generator import ArticleGenerator
from article_generator.article import Article
from article_generator.paragraph import Paragraph
from langchain.chat_models import ChatOpenAI

llm = ChatOpenAI(model="gpt-3.5-turbo-16k", temperature=0.7, max_tokens=8000)
article = Article(title="", paragraphs=[Paragraph(index=1, header="")])
gen = ArticleGenerator(llm=llm, article=article)
print(gen.complete().to_markdown())

See main.py and sample.py for full examples.

Stack

• Python 3.10+
• LangChain
• OpenAI
• ChromaDB (optional, for resource grounding)

License

MIT