DEV.to Post Generator

This project is an AI-powered assistant for generating high-quality, engaging blog posts for DEV.to. It leverages multiple LLMs (via Ollama) for reasoning, content generation, validation, and image suggestion, and integrates with the DEV.to API for publishing.

Features

• Fetches your DEV.to articles and analyzes top-performing posts.
• Summarizes previous articles to extract themes, styles, and gaps.
• Generates new blog drafts in markdown, tailored for high engagement.
• Analyzes and validates generated content for novelty and factual accuracy.
• Suggests image keywords for visual enhancement.
• Human-in-the-loop options: review, edit, or publish drafts directly or as a draft on DEV.to.

Setup

1. Clone the repository and install dependencies:

   git clone https://github.com/Vikranth3140/dev.to-post-generator.git
   cd dev.to-post-generator
   pip install -r requirements.txt

2. Set up your environment variables:
   • Create a .env file in the project root:

     DEV_API_KEY=your_dev_to_api_key
     

   • Get your DEV.to API key here.
3. Start your Ollama server and ensure the required models (llama3.1, mistral, etc.) are available and running.

    ollama run llama3.1
    ollama run mistral

Usage

Run the main script:

python dev_post_generator.py

Follow the prompts to:

• Generate a new post draft based on your top articles.
• Review AI-generated analysis and fact-checking.
• Get suggested image keywords.
• Choose to publish, edit, redo, or save as a draft.

Architecture: Modular, Not Multi-Agentic

This project uses a modular architecture, where each function (reasoning, generation, validation, image suggestion) is invoked by a central orchestrator script. While these modules are referred to as "agents" for clarity, they are not autonomous agents in the sense of multi-agent systems—they do not operate independently, maintain their own state, or communicate with each other outside the main script.

Human-in-the-Loop Regulation

At each stage, the human user is prompted to:

• Review the generated draft, analysis, and fact-check results.
• Edit the draft with custom instructions, or request a complete redo.
• Approve and publish the post, or save it as a draft for further review.

This ensures that while AI modules automate content creation and validation, the final decision and oversight remain with the human user, providing a regulated, safe, and high-quality publishing workflow.

---

Note:
This project does not implement a true multi-agentic system as defined in AI or distributed systems research. All "agents" are stateless model calls orchestrated by a single script, with no inter-agent communication or autonomy.