Skip to navigation

Threat reportPhishingTL-2026-0832

GitBait: GitHub Pages (github.io) Phishing Campaign Targeting Mexican Banking & Financial Sector

highACTIVE

GitBait is a long-running, modular credential-phishing campaign that abuses GitHub Pages (github.io) to host pages impersonating Mexican banks. Active for roughly three years and continuously maintained since at least January 2025, it has targeted at least 12 local and foreign financial institutions operating in Mexico, exfiltrating harvested credentials primarily through the legitimate SheetBest API to attacker-controlled Google Sheets, with a Telegram-bot exfiltration path as an alternative.

Key facts

IdentifiersTL-2026-0832
Also known asGitBait
SeverityHIGH
Affected productsGitHub GitHub Pages (github.io); Mexican financial institutions Online banking / customer login portals (impersonated); SheetBest SheetBest API (api.sheetbest.com)
AttributionGitBait operators (low confidence)
MITRE ATT&CK techniquesT1027 Obfuscated Files or Information, T1036 Masquerading, T1041 Exfiltration Over C2 Channel, T1056 Input Capture, T1059 Command and Scripting Interpreter, T1071 Application Layer Protocol, T1102 Web Service, T1111 Multi-Factor Authentication Interception, T1566 Phishing, T1567 Exfiltration Over Web Service, T1583 Acquire Infrastructure, T1585 Establish Accounts and 3 more in the attack flow
MotivationFINANCIAL
Target sectorsfinancial, banking, fintech
Target regionsMexico, Latin America
Detection rules9 (1 shown free)
Sources7 cited sources

Detection coverage

As of 2026-06-17, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-0832 across Splunk SPL, Microsoft KQL and Sigma, covering 32 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.

Free detection: GitBait — SheetBest / Telegram / 159.89.254.93 Credential Exfiltration C2 (C2 / Exfiltration)

Detects GitBait credential exfiltration channels: browser POSTs to api.sheetbest.com/sheets/<uuid> (primary, abusing the legitimate Google-Sheets-to-REST API), POSTs to the Telegram Bot API (api.telegram.org/bot.../sendMessage secondary path), and any connection to the observed exfil host 159.89.254.93:443. These are living-off-trusted-services exfil paths that evade domain reputation.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1567.002, T1041, T1102, T1071.001
  • Data sources: proxy, secure_web_gateway, DeviceNetworkEvents, firewall
  • Severity: critical
  • Confidence: high

Splunk SPL

index=proxy sourcetype="proxy:web"
| eval u=lower(coalesce(url, dest_host))
| where (match(u, "api\.sheetbest\.com/sheets/") AND http_method="POST")
    OR match(u, "api\.telegram\.org/bot.*/(sendmessage|senddocument)")
    OR dest_ip="159.89.254.93"
| eval channel=case(match(u,"sheetbest"),"sheetbest_primary", match(u,"telegram"),"telegram_secondary", dest_ip="159.89.254.93","exfil_host_do", true(),"other")
| stats count min(_time) as first_seen max(_time) as last_seen values(url) as urls values(http_user_agent) as user_agents by src_ip user channel dest_ip
| convert ctime(first_seen) ctime(last_seen)
| sort - count

Microsoft KQL

let exfilHosts = dynamic(["api.sheetbest.com","api.telegram.org"]);
let exfilIp = "159.89.254.93";
DeviceNetworkEvents
| where Timestamp > ago(24h)
| where RemoteUrl has_any (exfilHosts) or RemoteIP == exfilIp
| where InitiatingProcessFileName in~ ("chrome.exe","msedge.exe","firefox.exe","brave.exe","opera.exe")
| extend Channel = case(RemoteUrl has "sheetbest", "sheetbest_primary", RemoteUrl has "telegram", "telegram_secondary", RemoteIP == exfilIp, "exfil_host_do", "other")
| project Timestamp, DeviceName, InitiatingProcessFileName, InitiatingProcessAccountName, RemoteUrl, RemoteIP, Channel
| summarize Count=count(), FirstSeen=min(Timestamp), LastSeen=max(Timestamp), Urls=make_set(RemoteUrl, 25) by DeviceName, InitiatingProcessAccountName, Channel
| sort by Count desc

Sigma

title: GitBait SheetBest Telegram and DigitalOcean Credential Exfiltration Channels
id: 8b5e3c92-3f40-4c9d-be23-0832c3d4e5f6
status: experimental
description: Detects GitBait credential exfiltration over trusted web services — POSTs to api.sheetbest.com/sheets/<uuid>, the Telegram Bot API secondary path, or connections to the observed exfil host 159.89.254.93.
references:
    - https://www.group-ib.com/blog/gitbait-phishing-mexico-banking-finance/
    - https://attack.mitre.org/techniques/T1567/002/
    - https://attack.mitre.org/techniques/T1102/
author: Threadlinqs Team
date: 2026-06-17
tags:
    - attack.exfiltration
    - attack.command_and_control
    - attack.t1567.002
    - attack.t1041
    - attack.t1102
    - attack.t1071.001
logsource:
    category: proxy
detection:
    selection_sheetbest:
        c-uri|contains: 'api.sheetbest.com/sheets/'
        cs-method: 'POST'
    selection_telegram:
        c-uri|re: 'api\.telegram\.org/bot.*/(sendMessage|sendDocument)'
    selection_ip:
        dst_ip: '159.89.254.93'
    condition: selection_sheetbest or selection_telegram or selection_ip
fields:
    - src_ip
    - c-uri
    - dst_ip
    - cs-user-agent
falsepositives:
    - Legitimate business use of SheetBest or Telegram bots from sanctioned applications (allowlist those service accounts/hosts)
level: high

Known false positives:

  • Sanctioned internal tools that legitimately use SheetBest or Telegram bot integrations from server/service accounts
  • Personal Telegram usage on BYOD endpoints

Tuning notes: Scope to end-user browser processes to avoid flagging sanctioned server-side SheetBest/Telegram integrations; allowlist approved service accounts and the IPs of legitimate integrations. Highest fidelity when the initiating process is a browser and the session also accessed a github.io lure path.

8 more detection rules (Splunk SPL, Microsoft KQL, Sigma) · unlock with TL Creds. Sign up free to unlock the detection rules or query them with the Threadlinqs MCP server (Purple)
  • GitBait — Browser Navigation to github.io Lure Host (Exploitation / Initial Access) · unlock with TL Creds
  • GitBait — Browser POST to SheetBest API From End-User Session (Post-Exploitation / Credential Capture) · unlock with TL Creds
  • GitBait — Telegram Bot API & DigitalOcean Exfil Host Beacon (C2 / Lateral Movement) · unlock with TL Creds
  • GitBait — Phishing Page Content Signature: noindex + OpenGraph + Bootstrap-alpha + Randomized Script Path (Exploitation) · unlock with TL Creds
  • GitBait — Form-Intercept fetch() Credential Capture & Fake-Verification OTP Toggle (Post-Exploitation) · unlock with TL Creds
  • GitBait — Hardcoded Telegram Bot Token Secondary Exfil in Page Content (C2 / Exfiltration) · unlock with TL Creds
  • GitBait — github.io Banking Support/Cancelacion Lure Access (Exploitation / Initial Access) · unlock with TL Creds
  • GitBait — Numeric-Key Credential/OTP Exfil POST Body (Post-Exploitation / Credential Capture) · unlock with TL Creds

Indicators of compromise

As of 2026-06-17, Threadlinqs Intelligence tracks 32 indicators of compromise for TL-2026-0832.

Sample indicators (3 of 32)

  • IP address: 159.89.254[.]93
  • Domain: soporte-index25[.]github[.]io
  • Domain: soporte-index09[.]github[.]io

Defanged, TLP:CLEAR. Full set of 32 indicators: unlock with TL Creds.

32 indicators of compromise · unlock with TL Creds. Compare plans

Technical analysis

GitBait is a phishing operation tracked by Group-IB that weaponizes the trust and free hosting of GitHub Pages to impersonate at least 12 Mexican financial institutions. Rather than registering and burning standalone phishing domains, the operators publish credential-harvesting sites as Jekyll-based GitHub Pages projects deployed through GitHub Actions, producing 100+ distinct github.io subdomains (e.g. soporte-index25.github.io, sntdr-soporte25.github.io, support-vh.github.io). Because github.io is a high-reputation domain that defenders cannot easily blocklist without breaking legitimate developer content, the pages routinely bypass domain-reputation and email-security controls.

Victims are lured via messaging-app links (WhatsApp, SMS, Telegram) using brand-themed lure paths such as /cancelacion (57 observed), /soporte (43), /mb1, /st1, /soporte-cancelacion and /stdr, landing on harvesting endpoints like /index.html, /login.html, /1023.html and /159enlace.html. Pages are built on Bootstrap 5.3.0-alpha3 with Kanit/Play Google Fonts and carry Open Graph metadata so the links render convincing previews inside messaging apps, plus a `<meta name="robots" content="noindex,nofollow">` directive to evade search-engine indexing and crawler-based discovery.

The credential-capture logic lives in externally loaded JavaScript referenced through deliberately randomized, high-entropy path structures (e.g. src="/Vm7LSL59GGpJ8EK-ew/atw1fpkSb6ai/b3o0AQ/EE4aPHI/MCy8") to frustrate static analysis and signature matching. A form with id="contact-form" intercepts submissions using e.preventDefault() and a fetch() POST, then dynamically swaps the login interface for a verification screen (toggling elements id="registro" and id="exito"), consistent with capture of one-time codes / MFA tokens in addition to primary credentials.

Exfiltration relies on the legitimate SheetBest REST service (api.sheetbest.com), which turns attacker-controlled Google Sheets into a write-capable JSON API. Stolen data is POSTed as JSON using numeric keys ('1','2','3') mapped to credential field values, landing directly in the operators' spreadsheets via endpoints such as https://api.sheetbest.com/sheets/0e2a1336-e971-496f-9eb2-cd8dcd25565c. A secondary path embeds hardcoded Telegram bot tokens and chat IDs in the JavaScript for real-time delivery. An exfiltration server at 159.89.254.93:443 was also observed.

Group-IB attributes the operation to a small, persistent crew: the GitHub repository shows 66 commits by 3 contributors, with the earliest operator activity on 2025-01-16 (account ss-soporte, rronromo@gmail.com) and the most recent on 2026-03-12 (SheetBest endpoint rotation). The accounts ce-soporte and soporte-s1 share the email jejrvsbdb@gmail.com, indicating at least one operator manages multiple accounts. The campaign is financially motivated credential theft (no CVE, no malware binary involved) and remains active, posing a direct fraud and account-takeover risk to Mexican banks and their customers.

MITRE ATT&CK techniques by tactic:

Defense Evasion

T1027 Obfuscated Files or Information; T1036 Masquerading

Exfiltration

T1041 Exfiltration Over C2 Channel; T1567 Exfiltration Over Web Service

Credential Access

T1056 Input Capture; T1111 Multi-Factor Authentication Interception

Collection

T1056 Input Capture

Execution

T1059 Command and Scripting Interpreter

Command and Control

T1071 Application Layer Protocol; T1102 Web Service

Initial Access

T1566 Phishing

Resource Development

T1583 Acquire Infrastructure; T1585 Establish Accounts; T1608 Stage Capabilities

Reconnaissance

T1598 Phishing for Information

stealth

T1684.001 Impersonation

Remediation

Immediate actions

  • Block and report the known github.io phishing subdomains and the exfiltration host 159.89.254.93 at perimeter/proxy
  • Alert on and block outbound POST requests to api.sheetbest.com from end-user browsing sessions
  • Submit takedown / abuse reports to GitHub for repositories impersonating your brand
  • Push customer-facing fraud warnings about fake 'soporte'/'cancelacion' support pages delivered via WhatsApp, SMS and Telegram

Workarounds

  • Filter inbound messaging-app links pointing to *.github.io support/cancellation paths
  • Use DNS/secure-web-gateway policy to inspect and restrict api.sheetbest.com from corporate endpoints

Longer-term hardening

  • Stand up continuous brand-monitoring for github.io subdomains using your institution's naming patterns (e.g. [brand]-soporte, [brand]-cancelacion)
  • Deploy phishing-resistant MFA (FIDO2/passkeys) to neutralize credential + OTP harvesting
  • Implement behavioral detection on credential-submission flows rather than relying on domain blocklists alone
  • Educate customers that legitimate banking support is never hosted on github.io

Affected and fixed versions

  • GitHub — GitHub Pages (github.io)
    Vulnerable versions: GitHub Pages hosting abused for phishing
  • Mexican financial institutions — Online banking / customer login portals (impersonated)
    Vulnerable versions: At least 12 local and foreign institutions operating in Mexico
  • SheetBest — SheetBest API (api.sheetbest.com)
    Vulnerable versions: Legitimate Google-Sheets-to-REST API abused for credential exfiltration

Timeline

  • Estimated campaign origin: Group-IB assesses GitBait has been operating for approximately three years prior to the June 2026 disclosure, predating the earliest preserved GitHub commit history (date approximate, derived from the ~3-year-active assessment).
  • Earliest preserved operator activity: GitHub account 'ss-soporte' (locked-ioc-25) performs initial repository setup and base infrastructure creation for the phishing kit.
  • Account 'ce-soporte' (locked-ioc-24) activates GitHub Pages hosting, bringing the credential-harvesting site online on a github.io subdomain.
  • Account 'soporte-s1' (locked-ioc-24, shared email with ce-soporte) adds new impersonated-institution templates and removes others, indicating a single operator running multiple accounts and active brand-targeting curation.
  • Account 'soporte-<BRAND>01' (locked-ioc-15) updates the credential-harvesting pages with brand-specific theming for a targeted Mexican bank.
  • Account '<BRAND>-soperte12' (locked-ioc-18) rotates the SheetBest credential-collection endpoint (reported as currently active) — the most recent observed operator activity, showing the operation remained live into 2026 across 66 commits by 3 contributors.
  • At disclosure, the campaign's documented scope comprised 100+ github.io phishing domains, 6 distinct SheetBest API exfiltration endpoints, and 7 primary lure-path variants (e.g. /cancelacion observed 57 times, /soporte 43 times), confirming long-running, continuously maintained operations.
  • Group-IB publicly discloses the GitBait campaign in a blog report, documenting the GitHub Pages abuse, SheetBest/Telegram exfiltration, and targeting of at least 12 Mexican financial institutions.

Sources cited

Previous and next in phishing

Drafted by the Threadlinqs research pipeline from 7 sources; published ; last updated —.

Threadlinqs Intelligence — Real-Time Threat Detection Platform

// threat_feed
$ sort --newest
Showing all threats

Live intelligence console

Threat level

Latest Threats