@detiq/crawler 1.1.7 → 1.1.9

package/dist/profiles/domain-profile.js

@@ -1,140 +1,135 @@
 const COMMON_REQUIRED_ARTIFACTS = [
-  "screenshot",
-  "dom",
-  "markdown",
-  "accessibility",
-  "network",
-  "performance"
+    'screenshot',
+    'dom',
+    'markdown',
+    'accessibility',
+    'network',
+    'performance',
 ];
 const PROFILES = {
-  ecommerce: {
-    industry: "ecommerce",
-    importantScreenPatterns: ["catalog", "product", "search", "filter", "cart", "checkout", "payment", "order", "returns"],
-    workflowGoals: ["discover product catalogue", "map cart and checkout", "capture order confirmation and failure states", "map search/filter/sort"],
-    piiRules: ["mask email", "mask phone", "mask address", "mask card-like numbers"],
-    riskyActions: ["place order", "submit payment", "save card", "apply real coupon"],
-    safeProbeRules: ["do not click final payment/order buttons", "use invalid test card data only", "stop before irreversible purchase"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  banking: {
-    industry: "banking",
-    importantScreenPatterns: ["login", "mfa", "dashboard", "accounts", "transactions", "transfer", "beneficiary", "statement", "dispute"],
-    workflowGoals: ["map account overview", "map transactions and statements", "map transfer setup without submitting", "map MFA/session timeout states"],
-    piiRules: ["mask account numbers", "mask balances", "mask names", "mask addresses", "mask tax identifiers"],
-    riskyActions: ["transfer money", "add beneficiary", "submit dispute", "download sensitive statement"],
-    safeProbeRules: ["read-only by default", "never submit transfer confirmation", "never alter beneficiary data"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  fintech: {
-    industry: "fintech",
-    importantScreenPatterns: ["kyc", "wallet", "transactions", "cards", "payments", "payouts", "risk", "compliance", "disputes"],
-    workflowGoals: ["map KYC onboarding", "map wallet/payment states", "map failed payment and dispute flows", "capture compliance screens"],
-    piiRules: ["mask card numbers", "mask bank accounts", "mask identity documents", "mask balances"],
-    riskyActions: ["send payment", "withdraw funds", "change card status", "submit KYC with real data"],
-    safeProbeRules: ["sandbox data only", "stop before money movement", "never upload real identity documents"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  healthcare: {
-    industry: "healthcare",
-    importantScreenPatterns: ["patient", "appointment", "claims", "provider", "prescription", "records", "lab", "consent"],
-    workflowGoals: ["map patient lookup safely", "map appointment booking before final submit", "map claims and prescription screens", "capture consent and records UX"],
-    piiRules: ["mask PHI", "mask patient names", "mask MRN", "mask DOB", "mask diagnosis", "mask prescriptions"],
-    riskyActions: ["schedule real appointment", "submit claim", "change medication", "message provider with PHI"],
-    safeProbeRules: ["use synthetic patients only", "do not submit clinical changes", "avoid external notifications"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  pharma: {
-    industry: "pharma",
-    importantScreenPatterns: ["product", "dosage", "safety", "batch", "adverse-event", "regulatory", "trial", "label"],
-    workflowGoals: ["map product and safety content", "map adverse event intake before submit", "capture regulatory and batch workflows"],
-    piiRules: ["mask patient reports", "mask reporter contact", "mask trial participant identifiers"],
-    riskyActions: ["submit adverse event", "change safety record", "publish regulatory content"],
-    safeProbeRules: ["stop before regulatory or safety submission", "use synthetic reports only"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  agriculture: {
-    industry: "agriculture",
-    importantScreenPatterns: ["crop", "field", "weather", "inventory", "marketplace", "logistics", "sensor", "subsidy"],
-    workflowGoals: ["map crop and field planning", "map inventory/marketplace flows", "capture logistics and sensor dashboards"],
-    piiRules: ["mask farmer identity", "mask location precision", "mask subsidy identifiers"],
-    riskyActions: ["place market order", "submit subsidy application", "change sensor/device control"],
-    safeProbeRules: ["avoid irreversible orders", "avoid device actuation", "use test farm data"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  "legacy-enterprise": {
-    industry: "legacy-enterprise",
-    importantScreenPatterns: ["frameset", "iframe", "popup", "table", "search", "admin", "report", "export"],
-    workflowGoals: ["map menu trees", "capture iframe and popup states", "map table filters and reports", "detect session timeout"],
-    piiRules: ["mask table cells that look like PII", "mask exported identifiers"],
-    riskyActions: ["bulk update", "delete record", "export production data", "submit admin change"],
-    safeProbeRules: ["read-only table probing", "never click delete/bulk actions", "capture popups without committing"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  saas: {
-    industry: "saas",
-    importantScreenPatterns: ["dashboard", "settings", "users", "billing", "integrations", "reports", "admin", "workspace"],
-    workflowGoals: ["map workspace navigation", "map settings and integrations", "map role/user management without changing state", "capture dashboards and reports"],
-    piiRules: ["mask emails", "mask tokens", "mask API keys", "mask billing data"],
-    riskyActions: ["delete workspace", "invite real user", "rotate key", "change billing plan"],
-    safeProbeRules: ["never submit destructive settings", "redact secrets immediately", "use sandbox users"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  },
-  unknown: {
-    industry: "unknown",
-    importantScreenPatterns: ["login", "dashboard", "settings", "search", "detail", "report", "form"],
-    workflowGoals: ["map top-level navigation", "capture forms and validation states", "discover hidden menus", "classify domain after first pages"],
-    piiRules: ["mask emails", "mask phone numbers", "mask long identifiers", "mask card-like numbers"],
-    riskyActions: ["delete", "submit payment", "transfer", "publish", "send notification"],
-    safeProbeRules: ["read-only by default", "stop before irreversible actions", "use synthetic data only"],
-    requiredArtifacts: COMMON_REQUIRED_ARTIFACTS
-  }
+    ecommerce: {
+        industry: 'ecommerce',
+        importantScreenPatterns: ['catalog', 'product', 'search', 'filter', 'cart', 'checkout', 'payment', 'order', 'returns'],
+        workflowGoals: ['discover product catalogue', 'map cart and checkout', 'capture order confirmation and failure states', 'map search/filter/sort'],
+        piiRules: ['mask email', 'mask phone', 'mask address', 'mask card-like numbers'],
+        riskyActions: ['place order', 'submit payment', 'save card', 'apply real coupon'],
+        safeProbeRules: ['do not click final payment/order buttons', 'use invalid test card data only', 'stop before irreversible purchase'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    banking: {
+        industry: 'banking',
+        importantScreenPatterns: ['login', 'mfa', 'dashboard', 'accounts', 'transactions', 'transfer', 'beneficiary', 'statement', 'dispute'],
+        workflowGoals: ['map account overview', 'map transactions and statements', 'map transfer setup without submitting', 'map MFA/session timeout states'],
+        piiRules: ['mask account numbers', 'mask balances', 'mask names', 'mask addresses', 'mask tax identifiers'],
+        riskyActions: ['transfer money', 'add beneficiary', 'submit dispute', 'download sensitive statement'],
+        safeProbeRules: ['read-only by default', 'never submit transfer confirmation', 'never alter beneficiary data'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    fintech: {
+        industry: 'fintech',
+        importantScreenPatterns: ['kyc', 'wallet', 'transactions', 'cards', 'payments', 'payouts', 'risk', 'compliance', 'disputes'],
+        workflowGoals: ['map KYC onboarding', 'map wallet/payment states', 'map failed payment and dispute flows', 'capture compliance screens'],
+        piiRules: ['mask card numbers', 'mask bank accounts', 'mask identity documents', 'mask balances'],
+        riskyActions: ['send payment', 'withdraw funds', 'change card status', 'submit KYC with real data'],
+        safeProbeRules: ['sandbox data only', 'stop before money movement', 'never upload real identity documents'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    healthcare: {
+        industry: 'healthcare',
+        importantScreenPatterns: ['patient', 'appointment', 'claims', 'provider', 'prescription', 'records', 'lab', 'consent'],
+        workflowGoals: ['map patient lookup safely', 'map appointment booking before final submit', 'map claims and prescription screens', 'capture consent and records UX'],
+        piiRules: ['mask PHI', 'mask patient names', 'mask MRN', 'mask DOB', 'mask diagnosis', 'mask prescriptions'],
+        riskyActions: ['schedule real appointment', 'submit claim', 'change medication', 'message provider with PHI'],
+        safeProbeRules: ['use synthetic patients only', 'do not submit clinical changes', 'avoid external notifications'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    pharma: {
+        industry: 'pharma',
+        importantScreenPatterns: ['product', 'dosage', 'safety', 'batch', 'adverse-event', 'regulatory', 'trial', 'label'],
+        workflowGoals: ['map product and safety content', 'map adverse event intake before submit', 'capture regulatory and batch workflows'],
+        piiRules: ['mask patient reports', 'mask reporter contact', 'mask trial participant identifiers'],
+        riskyActions: ['submit adverse event', 'change safety record', 'publish regulatory content'],
+        safeProbeRules: ['stop before regulatory or safety submission', 'use synthetic reports only'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    agriculture: {
+        industry: 'agriculture',
+        importantScreenPatterns: ['crop', 'field', 'weather', 'inventory', 'marketplace', 'logistics', 'sensor', 'subsidy'],
+        workflowGoals: ['map crop and field planning', 'map inventory/marketplace flows', 'capture logistics and sensor dashboards'],
+        piiRules: ['mask farmer identity', 'mask location precision', 'mask subsidy identifiers'],
+        riskyActions: ['place market order', 'submit subsidy application', 'change sensor/device control'],
+        safeProbeRules: ['avoid irreversible orders', 'avoid device actuation', 'use test farm data'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    'legacy-enterprise': {
+        industry: 'legacy-enterprise',
+        importantScreenPatterns: ['frameset', 'iframe', 'popup', 'table', 'search', 'admin', 'report', 'export'],
+        workflowGoals: ['map menu trees', 'capture iframe and popup states', 'map table filters and reports', 'detect session timeout'],
+        piiRules: ['mask table cells that look like PII', 'mask exported identifiers'],
+        riskyActions: ['bulk update', 'delete record', 'export production data', 'submit admin change'],
+        safeProbeRules: ['read-only table probing', 'never click delete/bulk actions', 'capture popups without committing'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    saas: {
+        industry: 'saas',
+        importantScreenPatterns: ['dashboard', 'settings', 'users', 'billing', 'integrations', 'reports', 'admin', 'workspace'],
+        workflowGoals: ['map workspace navigation', 'map settings and integrations', 'map role/user management without changing state', 'capture dashboards and reports'],
+        piiRules: ['mask emails', 'mask tokens', 'mask API keys', 'mask billing data'],
+        riskyActions: ['delete workspace', 'invite real user', 'rotate key', 'change billing plan'],
+        safeProbeRules: ['never submit destructive settings', 'redact secrets immediately', 'use sandbox users'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
+    unknown: {
+        industry: 'unknown',
+        importantScreenPatterns: ['login', 'dashboard', 'settings', 'search', 'detail', 'report', 'form'],
+        workflowGoals: ['map top-level navigation', 'capture forms and validation states', 'discover hidden menus', 'classify domain after first pages'],
+        piiRules: ['mask emails', 'mask phone numbers', 'mask long identifiers', 'mask card-like numbers'],
+        riskyActions: ['delete', 'submit payment', 'transfer', 'publish', 'send notification'],
+        safeProbeRules: ['read-only by default', 'stop before irreversible actions', 'use synthetic data only'],
+        requiredArtifacts: COMMON_REQUIRED_ARTIFACTS,
+    },
 };
 const INDUSTRY_HINTS = [
-  { industry: "ecommerce", terms: [/cart/i, /checkout/i, /product/i, /catalog/i, /sku/i, /wishlist/i, /order/i] },
-  { industry: "banking", terms: [/account/i, /statement/i, /beneficiar/i, /transfer/i, /iban/i, /routing/i, /balance/i] },
-  { industry: "fintech", terms: [/wallet/i, /kyc/i, /payout/i, /card/i, /payment/i, /risk/i, /dispute/i] },
-  { industry: "healthcare", terms: [/patient/i, /provider/i, /appointment/i, /claim/i, /prescription/i, /medical/i, /lab result/i] },
-  { industry: "pharma", terms: [/pharma/i, /dosage/i, /adverse/i, /batch/i, /regulatory/i, /clinical trial/i, /label/i] },
-  { industry: "agriculture", terms: [/crop/i, /field/i, /farm/i, /harvest/i, /sensor/i, /weather/i, /subsidy/i] },
-  { industry: "legacy-enterprise", terms: [/frameset/i, /iframe/i, /popup/i, /legacy/i, /report export/i, /mainframe/i] },
-  { industry: "saas", terms: [/workspace/i, /tenant/i, /integration/i, /billing/i, /api key/i, /admin/i, /dashboard/i] }
+    { industry: 'ecommerce', terms: [/cart/i, /checkout/i, /product/i, /catalog/i, /sku/i, /wishlist/i, /order/i] },
+    { industry: 'banking', terms: [/account/i, /statement/i, /beneficiar/i, /transfer/i, /iban/i, /routing/i, /balance/i] },
+    { industry: 'fintech', terms: [/wallet/i, /kyc/i, /payout/i, /card/i, /payment/i, /risk/i, /dispute/i] },
+    { industry: 'healthcare', terms: [/patient/i, /provider/i, /appointment/i, /claim/i, /prescription/i, /medical/i, /lab result/i] },
+    { industry: 'pharma', terms: [/pharma/i, /dosage/i, /adverse/i, /batch/i, /regulatory/i, /clinical trial/i, /label/i] },
+    { industry: 'agriculture', terms: [/crop/i, /field/i, /farm/i, /harvest/i, /sensor/i, /weather/i, /subsidy/i] },
+    { industry: 'legacy-enterprise', terms: [/frameset/i, /iframe/i, /popup/i, /legacy/i, /report export/i, /mainframe/i] },
+    { industry: 'saas', terms: [/workspace/i, /tenant/i, /integration/i, /billing/i, /api key/i, /admin/i, /dashboard/i] },
 ];
-function classifyDomainProfile(input) {
-  const haystack = [input.appUrl, input.projectName ?? "", input.appType ?? "", input.seedText ?? ""].join(" ");
-  const scores = /* @__PURE__ */ new Map();
-  for (const { industry, terms } of INDUSTRY_HINTS) {
-    for (const term of terms) {
-      if (term.test(haystack)) {
-        const prev = scores.get(industry) ?? { score: 0, evidence: [] };
-        prev.score += 1;
-        prev.evidence.push(term.source);
-        scores.set(industry, prev);
-      }
+export function classifyDomainProfile(input) {
+    const haystack = [input.appUrl, input.projectName ?? '', input.appType ?? '', input.seedText ?? ''].join(' ');
+    const scores = new Map();
+    for (const { industry, terms } of INDUSTRY_HINTS) {
+        for (const term of terms) {
+            if (term.test(haystack)) {
+                const prev = scores.get(industry) ?? { score: 0, evidence: [] };
+                prev.score += 1;
+                prev.evidence.push(term.source);
+                scores.set(industry, prev);
+            }
+        }
     }
-  }
-  let selected = "unknown";
-  let best = 0;
-  let evidence = [];
-  for (const [industry, result] of scores) {
-    if (result.score > best) {
-      selected = industry;
-      best = result.score;
-      evidence = result.evidence;
+    let selected = 'unknown';
+    let best = 0;
+    let evidence = [];
+    for (const [industry, result] of scores) {
+        if (result.score > best) {
+            selected = industry;
+            best = result.score;
+            evidence = result.evidence;
+        }
     }
-  }
-  const base = PROFILES[selected];
-  return {
-    ...base,
-    confidence: best === 0 ? 0.25 : Math.min(0.95, 0.35 + best * 0.15),
-    evidence
-  };
+    const base = PROFILES[selected];
+    return {
+        ...base,
+        confidence: best === 0 ? 0.25 : Math.min(0.95, 0.35 + best * 0.15),
+        evidence,
+    };
 }
-function getProfile(industry) {
-  const base = PROFILES[industry] ?? PROFILES.unknown;
-  return { ...base, confidence: 1, evidence: ["explicit"] };
+export function getProfile(industry) {
+    const base = PROFILES[industry] ?? PROFILES.unknown;
+    return { ...base, confidence: 1, evidence: ['explicit'] };
 }
-export {
-  classifyDomainProfile,
-  getProfile
-};
-//# sourceMappingURL=domain-profile.js.map