Tycoon2FA and the modern phish-kit landscape
TL;DR: Modern phishing-as-a-service (PhaaS) kits — Tycoon2FA (high-volume 2024–2025 against Microsoft 365), EvilProxy, Caffeine, NakedPages, ONNX Store, Sneaky 2FA, Strox — package AitM reverse-proxy infrastructure, captcha bypass, geo-IP filtering, and bot-detection evasion as turnkey services. Knowing the landscape helps you correlate detections to the kit and predict TTP evolution. Companion to aitm-evilginx-modern-phishing and phishing-infrastructure-design.
Why a kit landscape matters to defenders
- Attribution at “kit” level is easier than at “actor” level — many actors use the same kit.
- Defender detections become kit-specific — block the kit’s URL patterns, JavaScript signatures, captcha challenges.
- Kits update on a predictable cadence — knowing a kit got a new release tells you to refresh detections.
- Kit operators are sometimes traceable through their advertising channels (Telegram, dark-web markets).
Tycoon2FA
- Surfaced publicly in 2023; dominated 2024 telemetry.
- Targets primarily Microsoft 365.
- Operates as a hosted service — operator pays per-target.
- Notable features:
- Cloudflare-fronted infrastructure for resilience.
- Captcha bypass for automated targets (rotates between hCaptcha, reCAPTCHA, Cloudflare Turnstile).
- Geo-IP and ASN filtering to evade security researchers (won’t serve phishing to scanners or VPN IPs).
- JavaScript anti-debugging.
- Updated phishlets after Microsoft login UI changes.
- Phishlets supported: Microsoft 365, Outlook on the web, Adobe, DocuSign, Yahoo, AOL.
Sekoia, Proofpoint, and Trustwave have published detailed analyses.
EvilProxy
- Sold openly on dark-web markets since 2022.
- Built on customised Evilginx.
- Targets Google Workspace, Microsoft 365, Apple, Facebook, GitHub.
- Pricing tiered by target service.
- Known for fingerprinting and lying — different content served to security researcher IPs vs end-user IPs.
Resecurity wrote the original disclosure.
Sneaky 2FA / ONNX Store
- 2024-disclosed kits.
- Sneaky 2FA: distinct from Tycoon, sold to a different operator group.
- ONNX Store: targets corporate accounts; sold via Telegram.
- Use blob:// URLs and HTML attachments to evade email filters.
Caffeine
- 2022-discovered PhaaS by Mandiant.
- Self-service signup model (unlike most kits which require vetting).
- Phished M365 credentials at scale.
NakedPages
- 2023-disclosed.
- Targets primarily Microsoft 365 + Cloudflare-fronted.
Strox
- 2024.
- Distinct command-and-control infrastructure.
Common kit characteristics
- AitM core — see aitm-evilginx-modern-phishing.
- Captcha relay — proxies the IdP’s captcha so the user sees a legitimate-looking challenge.
- MFA prompt forwarding — push, number-matching, TOTP.
- Cookie capture — primary trophy; exfil to operator panel.
- Multi-target phishlets — one infrastructure serves several brand templates.
- Cloudflare / CloudFront fronting — content-delivery for performance + obfuscation.
- Telegram bot panel — operator interface.
- Subscription pricing — typically $200–$1000 / month.
Detecting kit traffic
Some signals are kit-specific; many are class-shared.
Network signals:
- Newly registered domain on Cloudflare with content proxying
login.microsoftonline.com-shaped HTML. - TLS fingerprint (JA3 / JA4) matching kit infrastructure.
- HTTP header sequence atypical for legitimate Microsoft endpoints.
Content signals:
- JavaScript with known kit-specific anti-debug strings.
- HTML structure with kit-specific class names or comments.
- Inline scripts that POST to non-Microsoft endpoints.
User-side signals:
- URL is not
login.microsoftonline.comor a documented internal AD FS endpoint. - Browser warning (rare; kits use valid certs).
Defensive playbook
- Phish-resistant MFA — FIDO2 / passkeys defeat every kit listed.
- Email + URL filtering — Microsoft Defender, Proofpoint, Mimecast, etc. updated for kit URL patterns.
- DNS filtering — Quad9 / Cloudflare Gateway / Cisco Umbrella block known-kit infrastructure.
- Block newly-registered domains at email gateway for first 30 days.
- Conditional Access — see conditional-access-bypass-modern.
- Defender for Cloud Apps — token theft alerts.
- Continuous Access Evaluation — invalidate captured tokens on risk events.
- Internal phishing simulation to train users.
Threat-intel sources for tracking kits
- Sekoia.io blog — Tycoon2FA tracking.
- Trustwave SpiderLabs blog.
- Microsoft Security blog — frequent AitM kit posts.
- Mandiant blog.
- Proofpoint — operational telemetry-driven reports.
- Group-IB — kit attribution.
- PhishTank, OpenPhish — URL feeds.
- DNSTwist alerts on lookalike registrations.
Operational notes for red teams
If contracted to test AitM susceptibility:
- Build infrastructure to match a known kit profile (so detections you tune actually catch the analogous adversary).
- Don’t reuse a kit’s panel — operate your own.
- Limit exfil — proof-of-concept token capture, immediately revoke.
Workflow to study (defender lab)
- Collect URL samples from PhishTank for the past 30 days.
- Categorise by phishing-kit signature (Tycoon, EvilProxy, etc.).
- For each kit, fingerprint network and content patterns.
- Author detection rules (Suricata / Zeek + email gateway).
- Test rules against a benign clone of the kit in lab.
Related
- aitm-evilginx-modern-phishing — underlying technique.
- phishing-infrastructure-design — your own infrastructure.
- mfa-fatigue-tradecraft — alternative MFA-defeat.
- oauth-device-code-phishing-m365 — non-kit alternative.
- domain-fronting-and-cdn-abuse — Cloudflare-fronting technique.