An automated, Web security audit tool that crawls a target website, runs security checks across 20+ scanner modules, enhances findings with Google Gemini AI, and presents everything in a real-time dashboard.

Live Demo: https://url-security-scanner.onrender.com

Security Scanner is a production-ready, full-stack security auditing platform built on:

Python + FastAPI — modular backend with isolated API router and WebSocket manager

Google Gemini AI (gemini-3.1-flash-lite-preview) — deep AI-powered analysis and remediation advice per finding

Real-time WebSocket streaming — live scan progress pushed to the browser without polling

Jinja2 templated HTML reports — self-contained, standalone scan reports rendered server-side

Scheduled auto-scan loop — continuous monitoring via a background daemon thread

The scanner performs both passive analysis (headers, cookies, DNS, JWT, SRI, cache control) and active probing (XSS, SQLi, SSRF, open redirect, path discovery, GraphQL introspection) against any target URL.

The dashboard has four pages accessible from the left sidebar:

Live metrics: Overall Risk, Total, Critical, High, Medium, Low, Info

AI-generated Executive Summary and Immediate Actions

Severity doughnut chart (Chart.js)

Live scan log preview (WebSocket streamed)

All findings from the latest scan or any loaded report

Filter bar — All / Critical / High / Medium / Low / Info

Each card shows: severity badge, title, affected location count, category

Expand (▼) to see: Description, Real-World Impact, Technical Details, numbered affected URLs with evidence snippets, Fix Suggestion, code example, OWASP tag, CWE tag

Last 20 scans sorted newest first

Shows target URL, time, severity badges, overall risk

Click any row to load findings into the Findings view

View Details ↗ opens the full standalone HTML report in a new tab

Delete removes both JSON + HTML report files

Full real-time log of every scanner step, pushed via WebSocket

Persists up to 200 lines · scroll-to-bottom on new entries

Control Description Target URL Website to scan Max Pages Crawler page limit Delay (s) Crawl delay between requests Hourly Auto-Scan Toggle background scan loop Run Scan Now Trigger an immediate scan

After all scanner modules run, raw findings are sent to Google Gemini in batches of 20:

Gemini enriches each finding with:

Field Description enhanced_severity Re-assessed severity based on real-world exploitability impact Plain-English real-world impact statement technical_details In-depth technical explanation fix_suggestion Specific, copy-paste-ready remediation code_example Working code fix (Nginx config, JS, response headers, etc.) priority Fix order (1 = most urgent) references CVE / CWE / OWASP references

Gemini also produces an executive summary with:

Overall risk level + risk score (0–100)

Top 3 key findings

Top 3 immediate actions

2–3 sentence executive summary

If Gemini is unavailable or rate-limited, the scanner falls back to raw findings gracefully.

Every completed scan writes two files to reports/:scan_<id>_<timestamp>.json

Full machine-readable report: all findings, occurrences, Gemini enhancements, severity counts, metadata.scan_<id>_<timestamp>.html

Standalone self-contained HTML file rendered from ui/templates/report_template.html via Jinja2:

Metric cards (Overall Risk, Total, Critical, High, Medium, Low, Info, Pages Scanned)

Executive Summary

Key Findings & Immediate Actions (two-column)

Interactive filterable findings with expandable detail panels

The scheduler (agent/scheduler.py) runs in a background daemon thread using the schedule library.

Toggle "Hourly Auto-Scan" in the sidebar UI, or

Set "hourly_scan_enabled": true in config.json

Runs a full scan every 60 minutes

Progress streams live to all connected WebSocket clients

New findings are merged into the existing history report for the same target — only net-new vulnerabilities are appended, avoiding duplicate noise

Auto-starts on server boot if enabled in config.json

Toggle off at any time from the sidebar.