# OpticParse & PhishVision -- Comprehensive AI Agent Integration Manifest (llms-full.txt)

> Standardized machine-readable reference for Autonomous LLM Agents, Agentic IDEs (Cursor, Windsurf, Devin), and Multi-Agent Orchestrators (LangChain, LlamaIndex, ElizaOS, AutoGPT, CrewAI).

---

## 1. Executive Summary & Capabilities

OpticParse and PhishVision provide a dual-engine architecture engineered for autonomous AI agents navigating the modern web:

1. **OpticParse (Visual Zero-CSS Web Scraping):**
   - Extracts clean, structured JSON from dynamic JavaScript-rendered web pages using multimodal edge vision.
   - Eliminates 96% of raw DOM/HTML token noise before context injection.
   - Bypasses anti-bot friction, dynamic layouts, and layout refactors without brittle CSS/XPath selectors.
2. **PhishVision (Real-Time Zero-Day Threat Shield):**
   - Sub-1.6s heuristic inspection for zero-day phishing kits, brand impersonation clones, and Web3 crypto wallet drainers.
   - Prevents autonomous agents from executing transactions or submitting credentials to compromised targets.

---

## 2. Fast Install & Free Trial Quota

All official client packages provide **200 free trial requests out of the box**. No credit card required.

### Python:
```bash
pip install langchain-opticparse
pip install llama-index-tools-opticparse
pip install opticparse-py
```

### TypeScript / Node.js:
```bash
npm install opticparse-agentkit-action
npm install opticparse-eliza-plugin
```

---

## 3. Production Code Recipes

### Recipe A: LangChain / Multi-Agent Swarm (Python)

```python
import os
from langchain_opticparse import OpticParseTool, PhishVisionTool

# Initialize tools (automatically adopts 200 free trial requests if key is not set)
optic_tool = OpticParseTool(api_key=os.getenv("OPTICPARSE_API_KEY", "op_live_trial_200"))
phish_tool = PhishVisionTool(api_key=os.getenv("OPTICPARSE_API_KEY", "op_live_trial_200"))

# Step 1: Verify target URL safety before agent navigation
threat_result = phish_tool.invoke({"url": "https://example-shop.com"})
print(f"Safety Verdict: {threat_result}")

# Step 2: Extract structured JSON via visual layout reasoning
scrape_result = optic_tool.invoke({
    "url": "https://example-shop.com/products",
    "query": "Extract list of items with name, price in USD, and in_stock status"
})
print(f"Structured Content: {scrape_result}")
```

### Recipe B: LlamaIndex Agent & RAG Pipeline (Python)

```python
import os
from llama_index.tools.opticparse import OpticParseToolSpec
from llama_index.core.agent import FunctionCallingAgentWorker

# Initialize tool specification
tool_spec = OpticParseToolSpec(
    api_key=os.getenv("OPTICPARSE_API_KEY", "op_live_trial_200")
)
tools = tool_spec.to_tool_list()

# Standalone execution
docs = tool_spec.extract(
    url="https://news.ycombinator.com",
    query="Extract top 5 news titles with external links and points"
)
print("Extracted Content:", docs[0].text)
```

### Recipe C: ElizaOS Autonomous Agent Swarm (TypeScript)

```typescript
import { opticParsePlugin } from "opticparse-eliza-plugin";

// Add opticParsePlugin directly to your ElizaOS Character definition
export const defaultCharacter = {
  name: "WebScout",
  plugins: [opticParsePlugin],
  settings: {
    secrets: {
      OPTICPARSE_API_KEY: process.env.OPTICPARSE_API_KEY || "op_live_trial_200"
    }
  }
};
// Available Actions:
// 1. VISION_SCRAPE: Visual extraction to structured JSON
// 2. PHISHVISION_DETECT: Threat inspection for malicious URLs
```

---

## 4. REST API Direct Endpoints

### Endpoint 1: Visual Extraction
- **URL:** `POST https://opticparse-mcp-portal.parastejpal987.workers.dev/mcp/tools/opticparse_extract`
- **Headers:** `Authorization: Bearer <API_KEY>`, `Content-Type: application/json`
- **Payload:**
  ```json
  {
    "url": "https://example.com/target",
    "query": "Extract pricing, title, description in JSON"
  }
  ```

### Endpoint 2: Zero-Day Threat Inspection
- **URL:** `POST https://opticparse-mcp-portal.parastejpal987.workers.dev/phishvision/scan`
- **Headers:** `Authorization: Bearer <API_KEY>`, `Content-Type: application/json`
- **Payload:**
  ```json
  {
    "url": "https://suspicious-domain.xyz"
  }
  ```

---

## 5. Monetization & Treasury Settlement (x402 Protocol)

When an agent or developer exhausts their 200 free trial requests, the API responds with `HTTP 402 Payment Required` or outputs an in-terminal QR code with instant recharge options:

1. **Credit Card / Stripe Portal:** Instant purchase of Starter ($29), Pro ($99), or Enterprise ($499) packs.
2. **On-Chain Micro-Payments (Zero-Gas):**
   - Supported Networks: **Base**, **Polygon**, **Arbitrum One**
   - Treasury Wallet: `0xd458E709e7d54fd3659EF66624A621Cde74EDD27`
   - Settlement Token: USDC / USDT
   - Verification: Pass transaction hash in header `X-Payment-TxHash` for instant automated crediting.

---

## 6. Official Registries & Links

- **Main Repository:** https://github.com/parastejpal987-cmyk/opticparse-public
- **Live Dual-Shield Space:** https://huggingface.co/spaces/paras9909/opticparse-vision-benchmark
- **Hugging Face Datasets:** https://huggingface.co/datasets/paras9909/opticparse-150-template-web-corpus
- **Glama MCP Registry:** https://glama.ai/mcp/servers/parastejpal987-cmyk/opticparse-public
- **Smithery Registry:** https://smithery.ai/server/@parastejpal987-cmyk/opticparse