Blog Generation Agent

An agentic blog generation system built with LangGraph and FastAPI. Given a topic, it autonomously generates a creative, SEO-friendly blog title and detailed Markdown content — with optional translation to Kannada or Japanese via conditional routing.

---

Architecture

The system uses LangGraph to define stateful, directed agent workflows. Two graph variants are supported depending on the API input:

Topic Graph (topic only)

START → title_creation → content_generation → END

Language Graph (topic + language)

START → title_creation → content_generation → route ──┬──> Kannada_translation → END
                                                       └──> Japanese_translation → END

---

Project Structure

Blog_Generation/
├── app.py                        # FastAPI application entry point
├── langgraph.json                # LangGraph Studio configuration
├── pyproject.toml                # Project dependencies (uv)
├── .env                          # Environment variables (not committed)
└── src/
    ├── graphs/
    │   └── graph_builder.py      # LangGraph graph definitions
    ├── nodes/
    │   └── blog_nodes.py         # Node logic: title, content, translation, routing
    ├── states/
    │   └── blogstate.py          # Typed state schema (BlogState, Blog)
    └── llms/
        └── openaillm.py          # OpenAI LLM wrapper (GPT-4o-mini)

---

Tech Stack

Component	Library / Tool
Agent Framework	LangGraph >= 1.1.3
LLM	OpenAI GPT-4o-mini via LangChain OpenAI
API Server	FastAPI + Uvicorn
State Schema	Pydantic v2 + TypedDict
Observability	LangSmith
Package Manager	uv
Python Version	>= 3.12

---

Setup

1. Clone the repository

git clone <repo-url>
cd Blog_Generation

2. Install dependencies

uv sync

3. Configure environment variables

Create a .env file in the project root:

OPENAI_API_KEY=your_openai_api_key
LANGCHAIN_API_KEY=your_langsmith_api_key

4. Run the server

python app.py

The API will be available at http://localhost:8002.

---

API Usage

POST /blogs

Generates a blog post from a topic, with optional language translation.

Request Body

Field	Type	Required	Description
topic	string	Yes	The subject for the blog post
current_language	string	No	Target language: "kannada" or "japanese"

Example — Topic only

curl -X POST http://localhost:8002/blogs \
  -H "Content-Type: application/json" \
  -d '{"topic": "The Future of Renewable Energy"}'

Example — With translation

curl -X POST http://localhost:8002/blogs \
  -H "Content-Type: application/json" \
  -d '{"topic": "The Future of Renewable Energy", "current_language": "kannada"}'

Response

{
  "data": {
    "topic": "The Future of Renewable Energy",
    "blog": {
      "title": "Harnessing Tomorrow: The Unstoppable Rise of Renewable Energy",
      "content": "## Introduction\n..."
    },
    "current_language": "kannada"
  }
}

---

Node Details

Node	Description
title_creation	Prompts the LLM to generate a creative, SEO-friendly blog title in Markdown
content_generation	Generates a detailed, structured Markdown blog post for the given topic
route	Passes current_language from state to enable conditional branching
route_decision	Conditional edge function — routes to Kannada_translation or Japanese_translation
Kannada_translation	Translates title and content to Kannada, preserving tone and formatting
Japanese_translation	Translates title and content to Japanese, preserving tone and formatting

---

LangGraph Studio

This project is configured for LangGraph Studio via langgraph.json. The blog_generator_agent graph (topic graph) is exposed for visual debugging and inspection.

langgraph dev

---

State Schema

class Blog(BaseModel):
    title: str    # Blog post title
    content: str  # Blog post body (Markdown)

class BlogState(TypedDict):
    topic: str              # Input topic
    blog: Blog              # Generated blog (title + content)
    current_language: str   # Target language for translation