Automate website quality assurance and bug detection using Weam’s AI-powered analysis workflow. Marketing agencies are flooded with website projects and face pressure to deliver market-competitive quality. Quality analysts follow extensive checklists covering functionality and UI, but manually reviewing entire websites for anomalies is time-consuming and exhausting. The QA Agent automates this process, helping teams identify issues quickly and maintain consistent quality standards.

​
What It Does

  • Crawl and analyze website URLs automatically
  • Detect bugs, errors, and quality issues across multiple dimensions
  • Generate detailed reports with actionable recommendations
  • Evaluate SEO, accessibility, security, and performance metrics
  • Provide insights for both functionality and UI improvements

​
How It Works

  1. Input – Submit website URL for analysis
  2. Crawling Process – Agent systematically reviews all website elements
  3. Analysis Engine – Evaluates against comprehensive quality checklist
  4. Output – Delivers detailed report covering SEO best practices, accessibility, code quality, security, content quality, and performance analytics

​
Why This Agent?

  • Eliminates repetitive manual QA processes
  • Ensures consistent quality standards across projects
  • Reduces time from hours to minutes for website analysis
  • Provides comprehensive coverage beyond human capacity
  • Helps beginners learn QA best practices

​
Built for Agencies – Adaptable for Any Web Team

Although designed for marketing agencies, this agent can be tailored for:
  • Development Teams – Pre-launch quality checks
  • E-commerce – Ensure optimal user experience
  • Enterprise IT – Maintain corporate website standards
  • Freelancers – Deliver professional-grade QA without a team

​
Overview

  • Goal: Automate website QA with AI + PageSpeed + scraping.
  • Input: Website URL
  • Output:
    • QA Checklist JSON
    • PageSpeed Analysis JSON
    • Styled Excel Report (via AWS S3)
    • Token usage and cost breakdown

​
Developer Guide :

​
AI (Python) — Execution Flow

AI Execution Steps:

1.  Initialize LLM
2.  Initialize Checklist & PageSpeed Queues
3.  Scrape the URL → Extract all code files (PHP, HTML, JS, CSS)
4.  Clean & Filter Unnecessary Code Lines
5.  Call PageSpeed API:
      ├─ Mobile Analysis
      └─ Desktop Analysis
6.  Count Tokens in Scraped Content
7.  Initialize LLM Memory & History
8.  Create LLM Chains:
      ├─ Scraped Content Analysis Chain
      └─ PageSpeed Checklist Analysis Chain
9.  Batch Checklist Items (10 per call)
10. Batch PageSpeed Results
11. Add Prompt Templates
12. Call LLM for Checklist Analysis (batch-wise)
      → Stream JSON Results into Checklist Queue
13. On Checklist Completion:
      → Start PageSpeed Analysis
      → Stream JSON Results into PageSpeed Queue
14. Parse PageSpeed JSON (Mobile & Desktop)
      → Format into UI-ready table structure
15. Combine All Results
      → Generate Final Excel Report
16. Upload Excel Report to AWS S3
17. Return Final API Response:
      ├─ PageSpeed JSON  
      ├─ Excel S3 Filename  
      └─ Token Usage & Cost Breakdown
18. Persist Results (Include in `additional_kwargs`)

​
AI (Python) — Core Components

🔧 Component💬 Description
scrape_url_contentScrapes all HTML, PHP, CSS, JS files from the provided URL
filter_code_linesCleans up unnecessary or duplicate lines for better analysis
pagespeed_api_callCalls Google PageSpeed API (Mobile + Desktop)
count_tokensEnsures token-safe LLM execution
initialize_llm_chainPrepares LLM chain for checklist and performance evaluation
batch_checklist_itemsGroups checklist items in chunks of 10
stream_to_queueStreams intermediate results into Redis Queue for real-time UI
parse_pagespeed_metricsExtracts useful metrics from PageSpeed JSON
create_excel_reportBuilds styled Excel report using all collected data
upload_to_s3Uploads the Excel file to AWS and returns the S3 path
calculate_costCalculates and logs token usage + pricing
store_responseSaves all results in DB for auditing or analytics

​
Next.js — Execution Flow

Node Execution Flow:

1  User selects "QA Specialist" agent from UI
2  Frontend Form Input → Sends Website URL to backend
3  Backend triggers Python pipeline via `proagent code`
4  Calls `getAIProAgentChatResponse()` to stream result
5  LLM streams results:
       └─ Live checklist items in frontend UI
6  Post-processing:
       ├─ QA Checklist rendered in Chat Window
       └─ Downloadable Excel link shown