Registre de canvis

May 18, 2026

• NOUelecterm dependency advisory check. GitHub repo scans can now flag npm manifests and lockfiles that pin or allow electerm versions affected by CVE-2026-41500 / GHSA-wxw2-rwmh-vr8f, with version-based advisory evidence rather than exploit confirmation.
• NOUOpenCms dependency advisory check. GitHub repo scans can now flag Maven pom.xml files that pin or resolve org.opencms:opencms-core versions affected by CVE-2023-42344 / GHSA-rcc6-6q2f-m2cw, with version-based advisory evidence rather than XXE exploit confirmation.

May 17, 2026

• NOUFUXA hardcoded JWT secret check. Verified active scans can now confirm CVE-2025-69971 exposure on FUXA instances that still trust the vulnerable fallback JWT signing configuration.
• NOUCKAN DataStore SQL exposure check. Verified active scans can now confirm unauthenticated CKAN DataStore SQL access associated with CVE-2026-42031 and guide teams to patched CKAN release lines or safer DataStore configuration.

16 May 2026

• NOUPDF.js dependency advisory check. GitHub repo scans can now flag npm manifests and lockfiles that pin or allow pdfjs-dist versions affected by CVE-2024-4367 / GHSA-wgrm-67xf-hhpq.
• NOUActive scans via REST API and MCP. Active scans can now be triggered from REST and MCP against verified domains that have been explicitly authorized from the dashboard. Authorization is revocable at any time.
• NOUSafer authorization levels for active scans. Domain authorization now distinguishes safer automated active checks from deeper active testing, so teams can automate the right level of verification for each domain.
• NOUFirst-use webhook for API/MCP active scans. A webhook can notify teams the first time an API/MCP-triggered active scan runs against a newly authorized domain.
• MILLORATImproved Referrer-Policy findings. Missing or weak Referrer-Policy results now separate URL-referrer leakage from broad information exposure, show document-response evidence, and include generic plus static-host remediation guidance.
• MILLORATImproved Permissions-Policy findings. Missing or weak Permissions-Policy results now show feature-level evidence, separate broad feature allowlists from missing hardening, and include generic plus static-host remediation guidance for common hosts, proxies, and app servers.
• MILLORATImproved clickjacking header prompts. Missing X-Frame-Options findings now point agents to CSP frame-ancestors as the modern protection, add Vercel/static SPA header guidance, and verify x-frame-options with CSP.
• MILLORATCSP header evidence and fix prompts improved. Missing-CSP reports now include clearer hosting and response context plus safer framework-aware remediation guidance.
• CORREGITVercel path-probe false positives reduced. FixVibe now requires stronger application-specific evidence before reporting exposed framework artifacts on deployments that rewrite unknown routes to the app shell.
• CORREGITCompliance findings no longer carry misleading CWE tags. The legal-compliance check previously tagged "missing privacy policy" and "missing terms of service" findings with CWE-359 (PII exposure), which doesn't describe the actual gap. Those findings now ship without a CWE — they're compliance/governance items, not classifiable security weaknesses.

May 15, 2026

• NOUAdditional research-informed checks. FixVibe shipped more coverage based on recent vulnerability research and mapped duplicate topics to existing scanner modules where coverage already existed.
• NOUVerificació de fuita de secrets al repositori. Les anàlisis de repositoris de GitHub ara poden marcar claus de proveïdor codificades de manera fixa i valors d'alta entropia similars a secrets enviats al codi, amb la prova emmascarada i la sol·licitud de rotació estàndard de FixVibe inclosa.
• NOUVercel deployment protection check. Passive scans can now flag public *.vercel.app generated deployment URLs that respond without Vercel Deployment Protection, while existing header checks continue to audit CSP, HSTS, and browser hardening.

May 14, 2026

• NOULiteLLM dependency advisory check. GitHub repo scans can now flag Python dependency files that pin or allow LiteLLM versions affected by CVE-2026-42208 / GHSA-r75f-5x8p-qvmc.
• NOULibreNMS dependency advisory check. GitHub repo scans can now flag Composer manifests that pin or allow LibreNMS versions affected by CVE-2024-51092 / GHSA-x645-6pf9-xwxw.
• MILLORATFirebase rules detection improved. BaaS scans now detect more Firebase app shapes and use read-only evidence to identify risky public data exposure.

May 13, 2026

• NOURepo Supabase RLS migration check. GitHub repo scans can now flag Supabase SQL migrations that create public tables without a matching ALTER TABLE ... ENABLE ROW LEVEL SECURITY statement.
• NOUSupabase Storage posture check. Passive scans can now review public Supabase Storage buckets and anonymous object-listing exposure alongside existing RLS and key checks.
• NOUAI-generated code guardrail check. GitHub repo scans can now flag missing security automation around code scanning, secret scanning, dependency updates, and AI-agent instructions.

May 12, 2026

• NOURepo web-app risk checklist. GitHub repo scans can now flag high-confidence OWASP-style code risks such as raw SQL interpolation, unsafe HTML sinks, credentialed wildcard CORS, disabled TLS verification, and weak JWT secret fallbacks.
• NOUNext.js middleware-bypass check. Active scans for verified domains can now confirm CVE-2025-29927 exposure on middleware-protected routes before reporting it, and reports include the standard FixVibe AI fix prompt for remediation.

9 de maig de 2026

• SEGURETATCross-origin scope hardening. Active scans and client-asset checks now stay within the authorized target scope and avoid carrying customer-provided credentials across cross-origin redirects.
• CORREGITSupabase RLS check is now strictly read-only. Supabase posture checks now avoid write attempts and focus on safe exposure signals. Verified-domain active testing remains the boundary for deeper confirmation.
• MILLORATLes troballes de capçaleres de seguretat només s’apliquen a respostes HTML arrel. La manca de CSP, Permissions-Policy, X-Frame-Options o Referrer-Policy en un 204, una API JSON, una descàrrega de fitxer o un 404 ja no produeix cap troballa. HSTS i X-Content-Type-Options encara es puntuen en totes les respostes.
• MILLORATAuth-flow and rate-limit checks now require stronger evidence. FixVibe now reports these issues only when the application behavior clearly supports the finding, reducing noise from generic error pages and unsupported methods.
• MILLORATFile-upload findings tier by exploitability evidence. File-upload reports now separate low-confidence acceptance signals from stronger evidence of risky serving behavior, reducing over-severity on benign upload handlers.

7 de maig de 2026

• CORREGITThreat-intel listing accuracy improved. FixVibe now distinguishes real blocklist evidence from resolver diagnostics so threat-intel findings do not over-report on infrastructure-side lookup responses.
• NOUEscaneigs de repositoris GitHub. Connecta un repo i FixVibe revisa el codi font per detectar claus de servei Supabase filtrades, tokens d’administració Firebase, fitxers de workflow arriscats i dependències desactualitzades, sense carregar mai el teu lloc desplegat. Consulta Tipus d’escaneig.
• NOUComprovacions SAST per JavaScript arriscat. Els escaneigs de repositori ara marquen new Function() i setTimeout("string"), tots dos equivalents a eval() quan reben entrada no fiable.
• CORREGITFalses troballes de “fitxer exposat” en llocs Vercel / Cloudflare. Les respostes nues 403 Forbidden ja no s’informen com a “el fitxer existeix”; la majoria de proveïdors edge retornen 403 per camins sospitosos tant si el fitxer hi és com si no. Ara exigim un senyal HTTP positiu abans de marcar-ho.
• CORREGITRepo-code false positives reduced. Repo scans now avoid flagging security terms in comments, documentation, test helpers, and clearly server-only contexts for several high-signal code checks.
• CORREGITLa clau anon de Supabase a localStorage ja no s’informa com una troballa de JWT en emmagatzematge; la clau anon és el token de client pensat per ser públic. Els tokens service-role reals en emmagatzematge del navegador ara són crítics i tenen un títol més clar.
• CORREGITCSP weakness detection improved. Content-Security-Policy checks now catch more permissive source policies while keeping evidence and remediation focused on the effective browser policy.
• CORREGITReflected-XSS check tightened. Active scans now require stronger reflection evidence before reporting executable-context risk, reducing false positives from unrelated markup on the page.
• CORREGITLa verificació de domini gestiona correctament redireccions apex ↔ www i és més clara sobre quin valor va al camp Host del registre TXT.

Estructura

Cada entrada està etiquetada perquè puguis escanejar-la ràpidament:

• NOU Una comprovació, superfície o funcionalitat nova.
• MILLORAT Un comportament existent ha millorat: més precís, més ràpid o més clar.
• CORREGIT Un error que vam publicar i després vam corregir.
• SEGURETAT Enduriment, correccions de vulnerabilitats o canvis de compliment.

Has detectat alguna cosa que s’ha trencat i no apareix aquí? Escriu a support@fixvibe.app.