# opticparse.com Listing Content

## OPTICPARSE — AI Vision Web Scraper
*(For existing listing — update these fields)*

### Short Description (100 chars)
Scrape any website with plain English. No CSS selectors. AI vision extracts structured JSON.

### Long Description
Opticparse is an AI-powered web scraping API that works fundamentally differently from traditional scrapers.

Instead of CSS selectors or XPath (which break every time a site updates), Opticparse:
- Launches a real Chromium browser with stealth mode enabled
- Navigates to your URL and waits for all JavaScript to load
- Screenshots the rendered page
- Sends the screenshot to a vision AI model
- Returns clean, structured JSON based on your natural language query

**Why this approach wins:**
✅ Works on React, Vue, Angular, and any JavaScript-heavy site
✅ Never breaks when a site redesigns their frontend
✅ No DOM inspection, no selector maintenance
✅ Anti-bot stealth mode bypasses basic Cloudflare detection
✅ Multi-provider AI (Groq, Vision AI (Groq LLaMA Vision + Gemini)) for fast, reliable responses

**Performance optimizations:**
- Blocks media/font/websocket assets — ~60% less bandwidth per request
- 5-minute response cache for repeated URL+query combinations
- AI provider rotation for zero downtime

### Tags
web-scraping, ai, vision, playwright, no-selector, data-extraction, python, fastapi, gpt4, automation

### Category
Data → Web Scraping

---

## PHISHVISION — AI Phishing & Prompt Injection Detector
*(New listing — use this content)*

### Short Description (100 chars)
Detect phishing pages & hidden AI prompt injection attacks using Vision AI (Groq LLaMA Vision + Gemini) vision analysis.

### Long Description
PhishVision is a cybersecurity API that catches what traditional phishing detectors miss — by actually rendering and analyzing webpages instead of just checking URL reputation.

**What it detects:**
1. **Brand impersonation phishing** — visual analysis identifies pixelated logos, wrong color schemes, and fake login forms even on brand-new domains not in any blacklist
2. **Hidden AI prompt injection** — extracts ALL page text including CSS-hidden elements to find attack strings like "IGNORE ALL PREVIOUS INSTRUCTIONS" targeting AI agents

**How it works:**
- Launches headless Chromium (bandwidth-optimized, blocks non-essential assets)
- Screenshots the rendered page in 1280×720
- Extracts `document.body.innerText` including hidden elements
- Sends both to Vision AI (Groq LLaMA Vision + Gemini) with a forensic analyst system prompt
- Returns structured JSON verdict

**Response schema:**
```json
{
  "verdict": "malicious | suspicious | safe",
  "confidence_score_percentage": 97,
  "impersonated_brand": "Microsoft | null",
  "threat_type": "brand_impersonation | prompt_injection | multiple | none",
  "visual_anomalies_detected": ["Pixelated logo", "Urgency message"],
  "hidden_payload_detected": "IGNORE ALL PREVIOUS INSTRUCTIONS... | null"
}
```

**Why PhishVision vs. VirusTotal/URLScan:**
- VirusTotal: checks URL reputation — useless for new domains
- URLScan: screenshots pages but doesn't analyze them
- PhishVision: AI forensic analysis of visual content AND hidden text

### Tags
cybersecurity, phishing, ai-security, prompt-injection, gpt4, vision, threat-detection, api

### Category  
Security → Threat Detection

### Pricing Tiers — BOTH APIS (Opticparse & PhishVision)

| Tier | Price | Calls/month | Features |
|---|---|---|---|
| **Free** | $0 | 50 req | Full access, all response fields — no credit card |
| **Pro** | $29 / mo | 2 000 req | Priority queue, no cold-start waits, email support |
| **Business** | $99 / mo | 10 000 req | Priority queue, SLA guarantee, dedicated support |
| **Enterprise** | $249 / mo | 50 000 req | Custom rate limits, SLA, white-glove onboarding |