addition of Phishing Email Detector Agent and case studies

[vaishnavisuvarnakar]

🛡️ Phishing Email Detector Agent

A security-focused sample agent that analyzes emails for phishing indicators
and provides risk scores with detailed red flag analysis.

---

🧠 What This Agent Does

Takes an email (demo, raw text, or real .eml file) and automatically:

1. Extracts structural features (sender domain, URLs, urgency language)
2. Scores it against known phishing patterns (0–100 risk score)
3. Runs LLM reasoning like a cybersecurity analyst
4. Outputs a clean report with red flags and actionable advice

---

⚙️ Core Changes

• sampleagents/phishing_detector_agent.py — main agent
• Our-Case-Study-to-understand_2.md — Traffic Agent case study
• Our-Case-Study-to-understand_3.md — Phishing Detector case study

DAG Structure: extract_features → check_patterns → llm_analyze → generate_report

Dual LLM Support:

• Google Gemini (gemini-1.5-flash-latest)
• OpenRouter (openrouter/free — auto-selects best free model)
• Simulated fallback if no API key — demo always works

Input Modes:

• Built-in demo emails (zero setup)
• --text raw string
• --eml email file
• --gemini / --openrouter for live LLM

---

▶️ How to Run

bash

Demo (no API key needed)

python sampleagents/phishing_detector_agent.py

With live LLM

python sampleagents/phishing_detector_agent.py --gemini YOUR_KEY

Real email file

python sampleagents/phishing_detector_agent.py --eml email.eml --gemini YOUR_KEY

screenshots of results attached

[vaishnavisuvarnakar]

@Precise-Goals that "merge branch 'Falcon-s-Hackathon:main' into main" commit came in my pr because i directly worked on main without making a branch. please ignore it.
P.S the email used for testing is that actual google AI bootcamp one from hack2skill