AI Web Scraping Markdown-to-JSON Parser

{
    "system_prompt": "You are a deterministic data normalization engine. You receive raw scraped markdown text from web pages. Strip all navigation bars, cookie banners, tracking scripts, and footers. Extract only the primary product or listing entities strictly as structured JSON.",
    "prompt_type": "automation_workflow",
    "framework": "n8n",
    "input_variables": [
        {
            "name": "scraped_markdown",
            "type": "string",
            "description": "Raw markdown or text extracted from web crawler"
        }
    ],
    "response_schema": {
        "type": "object",
        "properties": {
            "page_type": {
                "type": "string",
                "enum": [
                    "product_detail",
                    "listing_directory",
                    "blog_post",
                    "pricing_table"
                ]
            },
            "items": {
                "type": "array",
                "items": {
                    "type": "object",
                    "properties": {
                        "name": {
                            "type": "string"
                        },
                        "price": {
                            "type": "number"
                        },
                        "rating": {
                            "type": "number"
                        },
                        "availability": {
                            "type": "string"
                        },
                        "key_specs": {
                            "type": "array",
                            "items": {
                                "type": "string"
                            }
                        }
                    },
                    "required": [
                        "name"
                    ]
                }
            }
        },
        "required": [
            "page_type",
            "items"
        ]
    },
    "target_tools": [
        "n8n Firecrawl Node",
        "Puppeteer",
        "Stagehand"
    ],
    "compatible_models": [
        "GPT-4o",
        "Claude 3.5 Sonnet",
        "Gemini 2.5 Flash"
    ],
    "tags": [
        "n8n",
        "web-scraping",
        "firecrawl",
        "crawler",
        "json-parser",
        "ecommerce"
    ]
}