Phishing Risk Scoring System

I need a phishing-detection solution that scans both incoming Gmail messages and any links users encounter on the web, assigns each item a percentage-based risk score, and immediately sends an alert whenever the score crosses a defined threshold.

Scope of work
• Email analysis: Pull new messages through the Gmail API, extract headers, body text, and embedded URLs, then evaluate them with heuristics or machine-learning models for spoofed domains, look-alike characters, unusual senders, and malicious payloads.
• Webpage link scanning: Monitor links clicked or hovered over in the browser, perform real-time checks against threat-intel feeds and your model, and produce the same percentage score.
• Unified alerting: Once a link or email is flagged, push a notification (desktop, browser popup, or email) showing the risk percentage and the evidence that triggered it. No automatic quarantine or deletion—only the alert.
• Dashboard & logging: Simple web dashboard that lists recent scans, scores, and metadata for auditing and tuning.
• Extensibility: Structure the code so additional providers such as Outlook or Yahoo Mail can be added without a full rewrite.

Deliverables
1. Source code with clear inline comments.
2. Setup instructions and environment requirements.
3. Model or rule set used for scoring, plus a brief explanation of the weighting behind the percentage calculation.
4. Demo video or live walkthrough proving detection on sample phishing emails and malicious links.
5. Short write-up on how new indicators of compromise can be fed into the system.