langgraph-localai-tv

πŸ“Ί LangGraph LocalAI TV Pipeline

A three-stage news production pipeline built with LangGraph β€” designed for real workflows that branch, loop, wait for human input, and recover from errors.

πŸ— Architecture

β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”     β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”     β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  Stage 1        β”‚     β”‚  Human Approval   β”‚     β”‚  Stage 2           β”‚
β”‚  Fetch Headlines│────▢│  Gate (CLI/Auto)  │────▢│  Generate Raw Videoβ”‚
β”‚  (RSS/API)      β”‚     β”‚  βœ“ Approve/Reject β”‚     β”‚  (Mock/API)        β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜     β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜     β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                                                           β”‚
                        β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”               β”‚
                        β”‚  Stage 3         β”‚β—€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                        β”‚  Compile Bulletinβ”‚
                        β”‚  (JSON/Text)     β”‚
                        β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

✨ Features

πŸš€ Quick Start

1. Clone & Install

git clone https://github.com/nagarjunak-pixel/langgraph-localai-tv.git
cd langgraph-localai-tv
pip install -r requirements.txt

2. Configure

cp .env.example .env
# Edit .env with your API keys (optional β€” works without them)

3. Run

# Interactive mode (with human approval)
python pipeline.py

# Auto-approve mode (for demos/testing)
python pipeline.py --auto

# Resume a failed run
python pipeline.py --resume abc12345

πŸ“ Project Structure

langgraph-localai-tv/
β”œβ”€β”€ pipeline.py              # Main LangGraph pipeline runner
β”œβ”€β”€ human_approval.py        # Human-in-the-loop approval gate
β”œβ”€β”€ checkpoints.py           # SQLite checkpoint store
β”œβ”€β”€ nodes/
β”‚   β”œβ”€β”€ headline_fetcher.py  # RSS feed headline fetching
β”‚   β”œβ”€β”€ video_generator.py   # Video script + generation
β”‚   └── bulletin_compiler.py # Final bulletin assembly
β”œβ”€β”€ docs/                    # Documentation website (Netlify)
β”œβ”€β”€ requirements.txt
β”œβ”€β”€ .env.example
└── .gitignore

πŸ“– Documentation

**Live docs: https://langgraph-localai-tv.netlify.app

πŸŽ“ Learning Points

This project demonstrates:

  1. LangGraph StateGraph β€” Typed state flowing through pipeline nodes
  2. Checkpoint Persistence β€” Crash recovery via SQLite
  3. Human Approval Gates β€” Interactive CLI + auto-approve modes
  4. Error Recovery β€” Graceful degradation at each stage
  5. Modular Node Design β€” Each stage is independently testable
Project Framework Purpose
CrewAI Telugu Newsroom CrewAI Multi-agent news crew
Firecrawl NEXUS RSS Firecrawl Web scraping aggregator
DeerFlow Research Agent DeerFlow Long-horizon research

πŸ“„ License

MIT License β€” Built for Surya Agentic Academy


Author & Enterprise AI Architecture Hub

Built by Nagarjuna Reddy Kamireddy (Nagarjuna Kamireddy / Nagarjuna K) β€” Lead AI Architect, GenAI & Agentic AI Lead, and Corporate AI Trainer (20+ years across Enterprise Data, SAP HANA & Production AI; AWS Certified Solutions Architect – Professional & NVIDIA-Certified Professional: Agentic AI).