Skip to content

Latest commit

 

History

14 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🤖 Multi-Agent Content Engine

Python Version Framework LLM Platform License: MIT

An automated content marketing factory driven by a sequential crew of specialized AI agents. This application transforms any raw topic, engineering insight, or article link into highly optimized, platform-specific copy tailored for X (Twitter), LinkedIn, and Instagram instantly via a clean, unified web dashboard.


🔀 Multi-Branch Deployment Matrix

Choose the branch layout that best matches your target execution environment:

Branch Authentication Model Ideal Use Case Setup Required
main (Current) Bring-Your-Own-Key (UI) Public Serverless Cloud Hosting (Streamlit Community Cloud, Hugging Face) Zero backend files; Users paste their personal Gemini key into the secure sidebar text field.
feature-env-auth Environment Variable (.env) Local Private Testing, Private Servers, or Single-User Deployments Root level .env file containing hardcoded GEMINI_API_KEY credential values.

🚀 Features

  • Multi-Agent Collaboration: Harnesses an isolated, specialized workforce (Researcher $\rightarrow$ Copywriter $\rightarrow$ Editor) working sequentially instead of a single generalist LLM prompt.
  • High-Performance Cost Efficiency: Built natively using the Gemini 3.1 Flash-Lite core engine, providing high per-minute request limits (15 RPM) and rapid inference with 0$ overhead.
  • Production-Ready Web UI: Powered by Streamlit to offer users a frictionless, responsive dashboard completely divorced from the terminal environment.
  • Copy-Paste Optimized Outputs: Generates specialized layouts out of the box—including five-part X threads with conversion hooks, inline professional bulleted arrays for LinkedIn, and clean, punchy multi-tag descriptions for Instagram.

⚙️ How It Works

The engine uses a strict sequential execution assembly line model where data flows smoothly downstream, with each agent's output serving as the foundational context for the next task:

  [ User Input Topic ]
           │
           ▼
┌──────────────────────────────┐
│  1. Lead Content Researcher  │ ──► Extracts Top 3 Takeaways & Target Audience
└──────────────────────────────┘
           │
           ▼
┌──────────────────────────────┐
│  2. Social Media Copywriter  │ ──► Structures hooks, X Threads, & platform copy
└──────────────────────────────┘
           │
           ▼
┌──────────────────────────────┐
│    3. Chief Brand Editor     │ ──► Audits sentence flow, formatting, & line breaks
└──────────────────────────────┘
           │
           ▼
 [ Live Streamlit UI Display ]

🕵️‍♂️ Agent Workflow Deep-Dive

  • The Lead Content Researcher ingests your topic raw, parsing out fluff, verifying objective logical arguments, and classifying the primary target buyer persona or developer group.
  • The Social Media Copywriter steps in, accepting the researcher's breakdown as mandatory constraints, adapting voice frameworks, building engagement hooks, and implementing specific distribution angles for individual channels.
  • The Chief Brand Editor performs structural quality control, scrubbing alignment bugs, fine-tuning line breaks for optimal scannability, and ensuring execution blocks meet modern corporate publishing standards.

🛠️ Tech Stack

  • Orchestration Framework: CrewAI (Role-based agent design, automated cross-task memory handoffs)
  • Core Intelligence Engine: Google Gemini 3.1 Flash-Lite (Advanced multimodal text processing, highly optimized cost-to-performance scaling profile)
  • Storefront Web Framework: Streamlit (Sleek reactive Python dashboard state rendering, built-in session state variables mapping)

📦 Setup and Local Installation

Prerequisites

  • Python 3.12+ installed on an environment terminal path.
  • A valid Google AI Studio API Key.

1. Clone the Repository

git clone https://github.com/MohithReddy1/marketing-agent-crew.git
cd marketing-agent-crew

📦 2. Install Target Dependencies

Automatically build and compile the application's virtual container dependencies using the project requirements manifest:

pip install -r requirements.txt

🖥️ 3. Boot Up the Dashboard Storefront

Launch the local web server layer directly from your active interpreter environment context:

python -m streamlit run main.py

🔑 4. Provide Your Key via the UI

Once the dashboard renders inside your browser viewport, open the 🔑 Authentication Setup drawer in the left sidebar and paste your personal Gemini API key. The application state engine securely saves it for the remainder of your browser session!

🤝 Contributing & Future Roadmap

This project is fully open-source and free to use! Here is how you can get involved or scale it:

  • Developer Support: Feel free to fork this repository, open descriptive issues, or submit code optimization pull requests!
  • Custom Enterprise Adaptations: Looking to add automated live URL scraping or fine-tuned custom enterprise brand voice models? Feel free to reach out or drop a suggestion in the issues tab.

Distributed under the permissive MIT License. See the LICENSE file for complete rights and liability terms.