Merge pull request #2574 from trycompai/main

[comp] Production Deploy

Author: Marfuen

apps/api/src/trigger/vendor/vendor-risk-assessment-task.ts

@@ -476,7 +476,9 @@ export const vendorRiskAssessmentTask: Task<
     minTimeoutInMs: 1000,
     maxTimeoutInMs: 10000,
   },
-  maxDuration: 1000 * 60 * 10,
+  // 30 minutes total: Firecrawl Agent can take up to 25 min on slow SPA
+  // trust centers (Ubiquiti), and deep-scrape + DB writes need room too.
+  maxDuration: 1000 * 60 * 30,
   run: async (payload) => {
     await tags.add([`org:${payload.organizationId}`]);
 
@@ -1061,7 +1063,19 @@ export const vendorRiskAssessmentTask: Task<
           badgeCount: Array.isArray(complianceBadges)
             ? complianceBadges.length
             : 0,
+          complianceBadgesJson: JSON.stringify(complianceBadges ?? null),
           hasLogo: Boolean(logoUrl),
+          certificationsInAssessmentJson: JSON.stringify(
+            Array.isArray(
+              (coreData as { certifications?: unknown })?.certifications,
+            )
+              ? (
+                  coreData as {
+                    certifications?: Array<{ type: string; status: string }>;
+                  }
+                ).certifications
+              : [],
+          ),
         });
 
         // Update vendor with core data (keep status in_progress — news may still be loading)

apps/api/src/trigger/vendor/vendor-risk-assessment/deep-scrape-source-url.spec.ts

@@ -0,0 +1,147 @@
+import type { VendorRiskAssessmentCertification } from './agent-types';
+import { pickDeepScrapeSourceUrl } from './deep-scrape-source-url';
+
+const cert = (
+  overrides: Partial<VendorRiskAssessmentCertification> = {},
+): VendorRiskAssessmentCertification => ({
+  type: 'SOC 2 Type II',
+  status: 'verified',
+  issuedAt: null,
+  expiresAt: null,
+  url: null,
+  ...overrides,
+});
+
+describe('pickDeepScrapeSourceUrl', () => {
+  const vendorDomain = 'acme.com';
+
+  it("prefers 'Trust & Security' link over 'Security Overview'", () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [
+        { label: 'Security Overview', url: 'https://acme.com/security' },
+        { label: 'Trust & Security', url: 'https://acme.com/trust' },
+      ],
+      certifications: [],
+    });
+    expect(result).toBe('https://acme.com/trust');
+  });
+
+  it("falls back to 'Security Overview' when no 'Trust & Security' link", () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [{ label: 'Security Overview', url: 'https://acme.com/security' }],
+      certifications: [],
+    });
+    expect(result).toBe('https://acme.com/security');
+  });
+
+  it('falls back to a verified cert URL on the vendor domain when no labelled links match', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [],
+      certifications: [
+        cert({ url: 'https://acme.com/reports/soc2.pdf', status: 'verified' }),
+      ],
+    });
+    expect(result).toBe('https://acme.com/reports/soc2.pdf');
+  });
+
+  it('skips subdomain-matching cert URL when status is not verified', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [],
+      certifications: [
+        cert({ url: 'https://trust.acme.com/iso', status: 'unknown' }),
+      ],
+    });
+    expect(result).toBeNull();
+  });
+
+  it('accepts subdomain-matching cert URL (same registrable domain)', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [],
+      certifications: [
+        cert({ url: 'https://trust.acme.com/iso', status: 'verified' }),
+      ],
+    });
+    expect(result).toBe('https://trust.acme.com/iso');
+  });
+
+  it('rejects off-domain labelled links', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [
+        { label: 'Trust & Security', url: 'https://acme.trust.page' },
+      ],
+      certifications: [],
+    });
+    expect(result).toBeNull();
+  });
+
+  it('rejects off-domain verified cert URL', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [],
+      certifications: [
+        cert({ url: 'https://acme.safebase.io/soc2', status: 'verified' }),
+      ],
+    });
+    expect(result).toBeNull();
+  });
+
+  it('rejects unparseable URLs', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [{ label: 'Trust & Security', url: 'not a url' }],
+      certifications: [cert({ url: 'also not a url', status: 'verified' })],
+    });
+    expect(result).toBeNull();
+  });
+
+  it('returns null when everything is empty', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [],
+      certifications: [],
+    });
+    expect(result).toBeNull();
+  });
+
+  it('returns first verified cert URL and ignores later verified certs', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [],
+      certifications: [
+        cert({
+          type: 'SOC 2',
+          status: 'verified',
+          url: 'https://acme.com/first.pdf',
+        }),
+        cert({
+          type: 'ISO 27001',
+          status: 'verified',
+          url: 'https://acme.com/second.pdf',
+        }),
+      ],
+    });
+    expect(result).toBe('https://acme.com/first.pdf');
+  });
+
+  it('skips verified certs whose URL is null and continues to next cert', () => {
+    const result = pickDeepScrapeSourceUrl({
+      vendorDomain,
+      links: [],
+      certifications: [
+        cert({ type: 'SOC 2', status: 'verified', url: null }),
+        cert({
+          type: 'ISO 27001',
+          status: 'verified',
+          url: 'https://acme.com/iso.pdf',
+        }),
+      ],
+    });
+    expect(result).toBe('https://acme.com/iso.pdf');
+  });
+});

apps/api/src/trigger/vendor/vendor-risk-assessment/deep-scrape-source-url.ts

@@ -0,0 +1,47 @@
+import type { VendorRiskAssessmentCertification } from './agent-types';
+
+/**
+ * Resolve the best "source URL" to feed into `deepScrapeTrustPortal`.
+ *
+ * Fallback order:
+ *   1. The Agent-returned link labelled "Trust & Security" if it's on the vendor's domain.
+ *   2. The Agent-returned link labelled "Security Overview" if it's on the vendor's domain.
+ *   3. The URL of any verified certification that's on the vendor's domain.
+ *
+ * Returns null if nothing qualifies. Off-domain URLs are rejected at every tier —
+ * `deepScrapeTrustPortal` applies an additional third-party-portal gate, but
+ * this helper is the first line of defense against scraping an unrelated host.
+ */
+export function pickDeepScrapeSourceUrl(args: {
+  vendorDomain: string;
+  links: Array<{ label: string; url: string }>;
+  certifications: VendorRiskAssessmentCertification[];
+}): string | null {
+  const { vendorDomain, links, certifications } = args;
+
+  const isOnVendorDomain = (url: string): boolean => {
+    try {
+      const host = new URL(url).hostname.toLowerCase();
+      return host === vendorDomain || host.endsWith(`.${vendorDomain}`);
+    } catch {
+      return false;
+    }
+  };
+
+  const byLabel = (label: string) =>
+    links.find((l) => l.label === label && isOnVendorDomain(l.url))?.url ??
+    null;
+
+  const trustUrl = byLabel('Trust & Security');
+  if (trustUrl) return trustUrl;
+
+  const securityUrl = byLabel('Security Overview');
+  if (securityUrl) return securityUrl;
+
+  for (const cert of certifications) {
+    if (cert.status !== 'verified') continue;
+    if (cert.url && isOnVendorDomain(cert.url)) return cert.url;
+  }
+
+  return null;
+}