echo agent

Inspiration

Developers and content creators often juggle multiple platforms—drafting tailored social posts on one end while quick-testing and debugging code on the other. We created EchoAgent to unify social media management and developer utilities into a single, seamless AI workspace powered by Gemini.

What it does

EchoAgent acts as an all-in-one AI assistant workspace featuring two core modules:

AI Social Media Creator: Generates optimized posts for platforms like LinkedIn, Twitter/X, and Instagram. Users can customize tones (Professional, Casual & Friendly, Excited & Bold, Informative) and perform iterative refinements on the output.
Python Code Corrector: A lightweight browser workspace where users paste raw or broken Python code. Gemini analyzes the script, fixes syntax and logic errors (such as converting pseudocode into runnable turtle graphics), and delivers a detailed breakdown of what went wrong.
System Monitoring: Keeps track of real-time workspace health, uptime, and interaction history.

How we built it

Frontend: Next.js, HTML/CSS, and JavaScript for an interactive workspace interface.
AI Integration: Google Gemini API (@google/genai) for handling both conversational post generation/refinement and deep code syntax error analysis.
Backend & Hosting: Python backend APIs deployed on Vercel (https://nn-mu-azure.vercel.app/).

Challenges we ran into

Structured Code Analysis: Prompting Gemini to return both clean, runnable Python code and a distinct bug breakdown without breaking the UI layout.
Context Preservation: Ensuring tone adjustments and iterative refinements retained the original user context across multiple post-generation requests.

Accomplishments that we're proud of

Successfully building an dynamic Python syntax fixer that correctly interprets ambiguous pseudocode (like turning show a square into structured turtle code) and explains the fixes clearly.
Designing a clean, responsive single-dashboard UI that seamlessly combines marketing tools with developer utilities.

What we learned

Effective prompt engineering strategies for splitting complex tasks (code generation vs. diagnostic feedback).
Best practices for handling asynchronous AI state updates in real-time web workspaces.

What's next for EchoAgent

Direct Social Publishing: Integrating official APIs to post generated content directly to LinkedIn, Twitter/X, and Instagram.
Live Sandbox Execution: Integrating an in-browser WebAssembly/Pyodide runner so fixed Python code can execute directly on the dashboard.
RAG-based Context: Allowing users to connect personal documentation or previous posts for even tighter brand voice consistency.