@Euge

Devlog #7 - Phase 1 complete & Gemini prompt testing
Today I finished Phase 1 by testing valid, invalid and slow URLs and fixing client-side URL validation to prevent unnecessary PageSpeed API requests.
I also started Phase 2: designed and tested the Gemini prompt for generating explanation, why_it_matters and how_to_fix for each audit issue. The first AI outputs are working well, so I’m moving on to integrating this into the actual audit pipeline.

Hello guys! Welcome to my 6th devlog! 🔥🙃
Today I finally made a big improvement to the frontend. I got the initial idea from the internet and combined it with my own design skills and AI to create something that fits SiteScope AI.
I also finally figured out why I kept getting the “PageSpeed API daily limit reached” error, even though my 25,000-token quota was still available. Turns out I was running the backend from the wrong directory. 😭
I also started working on making the frontend more polished and attractive.
That’s all for today! The frontend actually took much longer than the 2 hours shown here, since I didn’t log some of yesterday’s and the day before’s work.
Btw, I’d really appreciate it if you checked out the video attached to this devlog! ❤️

DEVLOG #5 - Backend Fixes & Frontend
Today I worked mainly on the backend, fixed several issues with the Google PageSpeed API, solved the API quota problem and added CORS so the frontend can communicate with the backend properly.
I also started working on a better-looking frontend that should make SiteScope AI more attractive and easier to use… thanks for reading this devlog and watching that video.

DEVLOG #4 - Connecting the Frontend 🚀
Today I spent about 30 minutes finishing the basic frontend structure and connecting it to the backend.
What I did today:
● Connected script.js to the audit form
● Added a request from the frontend to /api/audit
● Added loading and error states
● Connected the frontend to the backend
● Deployed the frontend to Netlify
The basic flow is now there:
URL → Frontend → Backend → PageSpeed API → Results
There is still a problem with the deployed version — currently I get:
Error: Failed to fetch
I haven’t fully solved the deployment/API setup yet, so the audit doesn’t work properly from the public website for now. If you are a bit interested in my project here is the url: sitescope-ai.netlify.app, thanks for visiting💗🔥, Eugene

DEVLOG #3 — Parsing, Error Handling & First Frontend
Today I cleaned up the PageSpeed response and added basic error handling.
The backend now:
● Extracts scores for Performance, SEO, Accessibility and Best Practices
● Lists detected issues with their title, description and severity
● Handles invalid URLs, timeouts and API errors
● Returns a simpler JSON response
● Works with the /api/audit endpoint
Example response:

{
  "url": "https://example.com",
  "scores": {
    "performance": 100,
    "accessibility": 96,
    "best-practices": 96,
    "seo": 80
  },
  "opportunities": []
}

I also started working on the frontend. The URL form and some basic CSS are now in place, including the Space Grotesk font.
Next step: connect the form to the backend and display the audit results in the browser.

DEVLOG #2 - First Working Audit
Today I continued working on SiteScope AI and got the first real website audit working.
The main goal was to connect the backend to Google PageSpeed Insights and turn its response into something easier to use.
What I worked on today:

Created the basic FastAPI backend
Added the GET / endpoint
Created the POST /api/audit endpoint
Connected the backend to the PageSpeed Insights API
Tested the PageSpeed API manually and explored its JSON response
Created a parser to simplify the PageSpeed response
Extracted Performance, SEO, Accessibility and Best Practices scores
Added the main opportunities/issues to the response
Created mock data and tested the parser
Tested the complete /api/audit endpoint with curl
Refactored the PageSpeed parser into a separate module
Fixed a bug with query parameter serialization

The endpoint can now receive a website URL and return its Lighthouse scores together with the main issues found on the website.
For example, when testing it with example.com, the API returned the scores and detected issues such as a missing meta description, non-descriptive link text, a missing main landmark and a missing charset declaration.
This is still only the backend foundation, but the main idea is finally starting to work.
Next, I want to continue building the analysis pipeline and eventually use AI to turn the raw audit data into useful explanations and recommendations.
Day 2.

DEVLOG #1 - Starting SiteScope AI 🚀
Today I discovered Stardance by Hack Club, and it finally gave me the motivation to start a project I’ve wanted to build for a while: SiteScope AI.
The idea is to build an AI-powered website auditor that can analyze websites and provide useful insights and recommendations.
Before starting, I used AI to help me understand the basic architecture, technologies and components I would need.
Then I got to work.
What I did today:
Created the GitHub repository
Added a README.md and MIT License
Cloned the repository locally
Connected the project to Hack Club Stardance
Registered it as a Software -> Websites project
Created the initial project structure
Added .gitignore and .env
Created a Python virtual environment
Split the project into backend and frontend
Added the required API keys:
PAGESPEED_API_KEY
GEMINI_API_KEY

Tested the API keys to make sure they work
Fixed the Python interpreter and virtual environment configuration

It’s still just the beginning, but the foundation is now ready.
Next step: start building the actual website analysis pipeline. 🔥
This is only Day 1.