Skip to content

Response, New York. Noah Park: Check Point Patches Actively Exploited Management Server Path Traversal. Detection, Dallas. Ava Okello: Report Finds Nearly Half of Deployed Detections Need Attention Before SOCs Can Trust Them.

SOCtember

Always First. Fast SOC News.

LevelBlue SpiderLabs screenshot of a Wazza phishing page spoofing Adobe Document Cloud, with a "Confirm to open document" panel showing a verification code, a Copy code button and an Open Document button, beside a blurred shared PDF.
Screenshot: LevelBlue SpiderLabs, a Wazza device code phishing landing page spoofing Adobe Acrobat, Oct 20, 2026 (fair use).

Detection

LevelBlue Says the New Wazza Phishing Kit Hides Microsoft 365 Device Code Theft Behind PDF Lures From Hijacked Mailboxes, Google Sites Wrappers and Rotating Decoy Company Websites

LevelBlue SpiderLabs researchers say the kit, active since at least early July, walks victims from a SharePoint-branded PDF through gated routing layers and anti-bot checks to a fake Adobe, DocuSign or OneDrive page that hands them a real Microsoft device code, and they published URL paths and API endpoints defenders can hunt for.

Ava Okello, DetectionLondon5 min read

LONDON - A phishing kit that LevelBlue SpiderLabs calls Wazza is stealing access to Microsoft 365 accounts without ever asking for a password, according to research the company published on Oct. 20. Instead, the kit shows victims a genuine Microsoft device code and talks them into entering it, which links the attacker's session to their account.

SpiderLabs researchers Cris Tomboc and Kevin Adriano said the campaign has been active since at least early July 2026 and was still running when they published. Their analysis follows earlier reporting by The Hacker News and traces how the kit has changed across waves.

From a hijacked mailbox to a fake document portal

The emails come from compromised accounts and carry PDF attachments with business-themed lures about invoices, contracts, confidential client documents and urgent review requests. The PDFs consist mostly of embedded images with Microsoft SharePoint branding, overlaid with hyperlink annotations. Some file names include the name of the compromised sender's company.

LevelBlue SpiderLabs diagram of the Wazza attack chain in seven steps: phishing email from a compromised account, PDF lure, redirection and staging links, the Wazza preloader that filters traffic to a decoy website, the device code phishing page, victim sign-in on a legitimate Microsoft page, and account compromise when the attacker obtains OAuth tokens.
Diagram: LevelBlue SpiderLabs, Wazza device code phishing campaign attack chain, Oct 20, 2026 (fair use).

SpiderLabs tracked two delivery chains. In one, the PDF links to a Google Sites page that uses Google's own embedded-content feature to load the attacker's page, storing the malicious address in a data-url attribute so it never appears in the iframe's source. In the other, the PDF links directly to .eu domains, sometimes through URL rewriting services and other trusted platforms used as redirectors.

LevelBlue SpiderLabs screenshots of four PDF lure documents that use Microsoft SharePoint and Microsoft 365 branding with a single button inviting the reader to open a shared file.
Screenshot: LevelBlue SpiderLabs, sample PDF lures using Microsoft SharePoint branding, Oct 20, 2026 (fair use).Sender and file details are redacted by LevelBlue.

Gates, decoys and anti-bot checks

The core of the kit is a 56 KB script that picks a theme, checks routing rules, shows a short "validating" animation and mints a victim-specific token before redirecting the browser to a routing subdomain named check, id or gate, later renamed portal, hub and svc. The configuration LevelBlue examined held 19 themes spanning file sharing, banking, payments, human resources, cloud infrastructure and e-signature brands.

Visitors who arrive the wrong way see something else. A request to the bare domain, or one that fails the subdomain checks, is routed to a fully built corporate website. The kit defines six such personas, including a private bank, a clinic, a law practice and an energy company, and switches between them every hour. "As a result, the same infrastructure can present different legitimate-looking organizations at different times," the researchers wrote.

LevelBlue SpiderLabs screenshots of six Wazza decoy corporate homepages for a private bank, a medical clinic, a legal counsel firm, a renewable energy company, a residential property firm and a project management software company.
Screenshot: LevelBlue SpiderLabs, landing pages of the Wazza decoy websites, Oct 20, 2026 (fair use).LevelBlue said the decoys were gone in the campaign's second wave.

A further layer interrogates the browser before the phishing page loads. Scripts look for signs of Selenium, WebDriver, Puppeteer, Playwright, PhantomJS and Nightmare automation, headless user agents and open developer tools, collect WebGL details, and encrypt the results with PBKDF2-SHA256 at 100,000 iterations and AES-256-GCM.

The device code handoff

At the final stage, the page asks its backend for a code. In LevelBlue's test the backend answered in about nine seconds with the verification code, its expiry time and a login.microsoftonline.com device authorization address. The page then polls every two seconds; when a code expires it fetches a new one, and when the backend reports the sign-in as used, it shows an "Access Verified" screen. LevelBlue also found alternate templates impersonating OneDrive and DocuSign, and said new service values showed the kit "continues to evolve."

What defenders should do

  • Disable the device code flow in Microsoft Entra ID where it is not needed, or use Conditional Access to restrict the device code grant to approved users, managed devices or trusted locations, as LevelBlue recommends.
  • Correlate email, web and identity logs: a device code sign-in shortly after a user opens a PDF attachment or an unfamiliar document-sharing page should be investigated as one chain.
  • Search proxy and URL logs for the parameters tok, tgm, m, tg, thm and ch, for the paths /onlsbdmn, /onfqdn, /eprfx and /prfxmsmach, and for /r?key= and /meline.
  • Monitor requests to /api/wazza-config, /api/mint-token, /api/fire-mint, /bb-landing and /api.php?svc=, and to unusually long or randomized workers[.]dev subdomains, including beacon-surge-sync[.]workers[.]dev and beaconoflinenew.lastfightlink.workers[.]dev.
  • Apply stricter handling to external PDF attachments that contain links, and tell users that entering an unexpected device code can hand an outside application access to their account.

Sources:


Ava Okello covers detection engineering and SOC operations for SOCtember from London.

Related stories

Detection desk