{
  "name": "Curb Appeal — Step 2: Selective Scrape",
  "nodes": [
    {
      "parameters": {},
      "id": "step2-trigger",
      "name": "Manual Trigger",
      "type": "n8n-nodes-base.manualTrigger",
      "typeVersion": 1,
      "position": [0, 0]
    },
    {
      "parameters": {
        "jsCode": "// Provide the homepage URL — the workflow discovers all key pages automatically\n// Slug is auto-generated from the domain name\nreturn [{\n  json: {\n    url: 'https://example.com'\n  }\n}];"
      },
      "id": "step2-config",
      "name": "Set Config",
      "type": "n8n-nodes-base.code",
      "typeVersion": 2,
      "position": [220, 0]
    },
    {
      "parameters": {
        "method": "POST",
        "url": "https://api.firecrawl.dev/v1/map",
        "sendHeaders": true,
        "headerParameters": {
          "parameters": [
            {
              "name": "Authorization",
              "value": "Bearer YOUR_FIRECRAWL_API_KEY"
            }
          ]
        },
        "sendBody": true,
        "contentType": "raw",
        "rawContentType": "application/json",
        "body": "={{ JSON.stringify({ url: $json.url, limit: 20 }) }}",
        "options": {}
      },
      "id": "step2-map",
      "name": "Firecrawl Map",
      "type": "n8n-nodes-base.httpRequest",
      "typeVersion": 4.2,
      "position": [440, 0]
    },
    {
      "parameters": {
        "jsCode": "const links = ($json.links || []);\nconst baseUrl = $('Set Config').first().json.url.replace(/\\/$/, '');\n\n// Parse hostname with regex — URL constructor not available in n8n sandbox\nconst hostnameMatch = baseUrl.match(/^https?:\\/\\/([^\\/]+)/);\nconst hostname = hostnameMatch ? hostnameMatch[1] : baseUrl;\nconst domain = hostname.replace(/^www\\./, '');\nconst slug = domain.replace(/\\.[^.]+$/, '').replace(/[^a-z0-9]+/gi, '-').toLowerCase();\n\nconst SKIP = /\\/(blog|post|article|news|press|legal|privacy|terms|sitemap|tag|category|feed|wp-|admin|login|404|cdn|assets|uploads|media)\\b/i;\nconst KEEP = /\\/(about|service|contact|menu|pricing|location|team|appointment|booking|offer|product|portfolio|gallery)\\b/i;\n\nconst filtered = [baseUrl];\n\nfor (const link of links) {\n  if (filtered.length >= 8) break;\n  const clean = link.replace(/\\/$/, '');\n  if (clean === baseUrl) continue;\n  if (SKIP.test(clean)) continue;\n  if (/sitemap/i.test(clean) || clean.match(/\\.[a-z]{2,4}$/i)) continue;\n  const pathMatch = clean.match(/^https?:\\/\\/[^\\/]+(\\/.*)?$/);\n  const path = pathMatch ? (pathMatch[1] || '/') : '/';\n  const depth = path.split('/').filter(Boolean).length;\n  if (depth <= 2 && (KEEP.test(clean) || depth === 1)) {\n    filtered.push(clean);\n  }\n}\n\nreturn [{ json: { filtered_urls: filtered, homepage: baseUrl, slug } }];"
      },
      "id": "step2-filter",
      "name": "Filter Key Pages",
      "type": "n8n-nodes-base.code",
      "typeVersion": 2,
      "position": [660, 0]
    },
    {
      "parameters": {
        "fieldToSplitOut": "filtered_urls",
        "options": {}
      },
      "id": "step2-split",
      "name": "Split URLs",
      "type": "n8n-nodes-base.splitOut",
      "typeVersion": 1,
      "position": [880, 0]
    },
    {
      "parameters": {
        "method": "POST",
        "url": "https://api.firecrawl.dev/v1/scrape",
        "sendHeaders": true,
        "headerParameters": {
          "parameters": [
            {
              "name": "Authorization",
              "value": "Bearer YOUR_FIRECRAWL_API_KEY"
            }
          ]
        },
        "sendBody": true,
        "contentType": "raw",
        "rawContentType": "application/json",
        "body": "={{ JSON.stringify({ url: $json.filtered_urls, formats: $json.filtered_urls === $('Filter Key Pages').first().json.homepage ? ['markdown', 'screenshot'] : ['markdown'] }) }}",
        "options": {}
      },
      "id": "step2-scrape",
      "name": "Firecrawl Scrape",
      "type": "n8n-nodes-base.httpRequest",
      "typeVersion": 4.2,
      "position": [1100, 0]
    },
    {
      "parameters": {
        "jsCode": "const data = $json.data || {};\nconst url = data.metadata?.sourceURL || '(unknown url)';\nconst markdown = data.markdown || '(no content scraped)';\nconst title = data.metadata?.title || url;\nconst screenshot = data.screenshot || null;\n\n// Extract real image URLs from markdown syntax before any filtering\nconst imgRegex = /!\\[.*?\\]\\((https?:\\/\\/[^)\\s]+)\\)/g;\nconst imageUrls = [];\nlet m;\nwhile ((m = imgRegex.exec(markdown)) !== null) {\n  const src = m[1];\n  if (\n    !src.endsWith('.svg') &&\n    !src.includes('data:') &&\n    !src.includes('pixel') &&\n    !src.includes('track') &&\n    !src.includes('maps.googleapis.com') &&\n    !src.includes('maps.gstatic.com') &&\n    !src.includes('google.com/maps')\n  ) {\n    imageUrls.push(src);\n  }\n}\n\nreturn [{ json: { formatted: `--- PAGE: ${title} | URL: ${url} ---\\n\\n${markdown}`, image_urls: imageUrls, screenshot } }];"
      },
      "id": "step2-format",
      "name": "Format Page",
      "type": "n8n-nodes-base.code",
      "typeVersion": 2,
      "position": [1320, 0]
    }
  ],
  "connections": {
    "Manual Trigger": {
      "main": [[{ "node": "Set Config", "type": "main", "index": 0 }]]
    },
    "Set Config": {
      "main": [[{ "node": "Firecrawl Map", "type": "main", "index": 0 }]]
    },
    "Firecrawl Map": {
      "main": [[{ "node": "Filter Key Pages", "type": "main", "index": 0 }]]
    },
    "Filter Key Pages": {
      "main": [[{ "node": "Split URLs", "type": "main", "index": 0 }]]
    },
    "Split URLs": {
      "main": [[{ "node": "Firecrawl Scrape", "type": "main", "index": 0 }]]
    },
    "Firecrawl Scrape": {
      "main": [[{ "node": "Format Page", "type": "main", "index": 0 }]]
    }
  },
  "pinData": {},
  "active": false,
  "settings": {
    "executionOrder": "v1"
  },
  "tags": []
}