<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Trapdex</title>
  <subtitle>The independent, live record of attacks that trick developers and employees into running something malicious, traced from lure to artifact to actor to infrastructure.</subtitle>
  <link href="https://intel.trapdex.com/feed.xml" rel="self"/>
  <link href="https://intel.trapdex.com/"/>
  <id>https://intel.trapdex.com/</id>
  <updated>2026-11-16T00:00:00Z</updated>

  <entry>
    <title>Optimizely Web project code altered to serve a crypto address swapper, card skimmer and iOS exploit kits</title>
    <link href="https://intel.trapdex.com/campaigns/optimizely-web-project-javascript-injection/"/>
    <id>https://intel.trapdex.com/campaigns/optimizely-web-project-javascript-injection/</id>
    <published>2026-11-16T00:00:00Z</published>
    <updated>2026-11-16T00:00:00Z</updated>
    <summary>On 16 Nov 2026 an attacker added code to the Project JavaScript of Optimizely Web projects, so Optimizely&#39;s own CDN served it to visitors of every affected site. Report URI found it in 188 of the 1,611 snippets it checked, belonging to 178 organisations. The code swapped crypto wallet addresses, skimmed cards typed into the sites&#39; own forms, sent cookies and screenshots to the attacker, and pushed iOS exploit kits to iPhones and iPads. Optimizely says an unauthorized party exploited a vulnerability in Experimentation, which it has fixed. Most snippets were cleaned the same evening.</summary>
  </entry>

  <entry>
    <title>Westlaw typosquat search ads lead legal staff to a fake sign-in page with a ClickFix paste-and-run prompt</title>
    <link href="https://intel.trapdex.com/campaigns/clickfix-westlaw-typosquat-bing-ads/"/>
    <id>https://intel.trapdex.com/campaigns/clickfix-westlaw-typosquat-bing-ads/</id>
    <published>2026-11-10T00:00:00Z</published>
    <updated>2026-11-10T00:00:00Z</updated>
    <summary>Paid Bing search ads for &#39;Westlaw&#39; sent users through a Ghost-hosted typosquat redirect to a fake Thomson Reuters sign-in page that copied a PowerShell command to the clipboard and told the visitor to run it. Huntress reported the cluster on 10 Nov 2026 as one of several ClickFix variants.</summary>
  </entry>

  <entry>
    <title>Binance USD 6.3B theft that began with a fake job interview and a homework repo</title>
    <link href="https://intel.trapdex.com/campaigns/binance-fake-interview-pyenv/"/>
    <id>https://intel.trapdex.com/campaigns/binance-fake-interview-pyenv/</id>
    <published>2026-09-01T00:00:00Z</published>
    <updated>2026-11-03T00:00:00Z</updated>
    <summary>According to Binance&#39;s post-mortem, a fake project director approached an engineer through a job site in Sep 2026. After staged HR and &#39;CEO&#39; interviews, the engineer ran a homework repo on a work Mac, and it replaced Homebrew&#39;s pyenv with a malicious copy. The attackers then moved through an internal AI tool to wallet-infrastructure staff and a deployment token, and drained funds through the signing service on 29 Oct 2026.</summary>
  </entry>

  <entry>
    <title>Trojanized anthropic.claude-code 2.1.312/2.1.313 served from Cursor&#39;s extension marketplace</title>
    <link href="https://intel.trapdex.com/campaigns/cursor-trojanized-claude-code-extension/"/>
    <id>https://intel.trapdex.com/campaigns/cursor-trojanized-claude-code-extension/</id>
    <published>2026-10-29T00:00:00Z</published>
    <updated>2026-10-30T00:00:00Z</updated>
    <summary>For about 11 hours on 29 Oct 2026, Cursor&#39;s marketplace served modified builds of Anthropic&#39;s official Claude Code extension under its real ID, reaching up to about 5,000 users. The loader collected host and git details, planted an attacker SSH key, turned on SSH, opened a reverse tunnel, set persistence and beaconed home. Microsoft&#39;s Marketplace, Open VSX and Anthropic&#39;s own systems were not affected.</summary>
  </entry>

  <entry>
    <title>Unattributed, 2026: a fake IT helpdesk on Teams talks staff into Quick Assist, then a Node.js implant or AD CS abuse</title>
    <link href="https://intel.trapdex.com/campaigns/teams-helpdesk-impersonation-quick-assist-2026/"/>
    <id>https://intel.trapdex.com/campaigns/teams-helpdesk-impersonation-quick-assist-2026/</id>
    <published>2026-09-02T00:00:00Z</published>
    <updated>2026-10-30T00:00:00Z</updated>
    <summary>Someone in an external Microsoft 365 tenant messages or calls an employee on Teams as IT or the helpdesk and talks them into Quick Assist or a Teams &#39;request control&#39; session. In the September 2026 cases the attacker then installed a malicious MSI that runs a JavaScript implant on a portable Node.js runtime and moved over WinRM toward domain controllers. In the October 2026 cases the attacker staged scripts on portable runtimes and abused AD CS (ESC1 or ESC8) to get certificates for privileged accounts. Microsoft did not attribute either report to an actor.</summary>
  </entry>

  <entry>
    <title>GhostAction 2026: fake &#39;security audit&#39; workflows via hijacked accounts, plus GitHub device-flow phishing</title>
    <link href="https://intel.trapdex.com/campaigns/ghostaction-security-audit-workflows/"/>
    <id>https://intel.trapdex.com/campaigns/ghostaction-security-audit-workflows/</id>
    <published>2026-08-31T00:00:00Z</published>
    <updated>2026-10-27T00:00:00Z</updated>
    <summary>With stolen tokens and credentials, the operator pushed fake security workflows straight to default branches to send CI secrets to a hardcoded IP. Activity grew from hundreds of repos in Aug-Sep to about 72k commits across about 23k repos between 13 and 21 Oct. In October the operator added a lookalike-GitHub phishing kit that relays password and 2FA and approves an OAuth device flow, and a stealer disguised as a font file that runs from a VS Code tasks.json.</summary>
  </entry>

  <entry>
    <title>LiteLLM 1.107.0: malicious PyPI wheels and Docker images via a hijacked release workflow</title>
    <link href="https://intel.trapdex.com/campaigns/litellm-1-107-0-release-workflow-hijack/"/>
    <id>https://intel.trapdex.com/campaigns/litellm-1-107-0-release-workflow-hijack/</id>
    <published>2026-10-10T00:00:00Z</published>
    <updated>2026-10-21T00:00:00Z</updated>
    <summary>A master key committed to a project repo in May was never revoked. In October it was used against an internal API that returned an org owner&#39;s GitHub token. The attacker exfiltrated CI secrets, bypassed branch protection and pointed the release workflow at a lookalike repo. All 1.107.0 wheels and the official Docker images, including latest tags, carried a .pth credential stealer; PyPI quarantined the release within two hours.</summary>
  </entry>

  <entry>
    <title>OVHcloud CLI GitHub release binaries and checksums swapped after an IDE-extension compromise</title>
    <link href="https://intel.trapdex.com/campaigns/ovhcloud-cli-release-asset-swap/"/>
    <id>https://intel.trapdex.com/campaigns/ovhcloud-cli-release-asset-swap/</id>
    <published>2026-10-19T00:00:00Z</published>
    <updated>2026-10-21T00:00:00Z</updated>
    <summary>A contributor&#39;s GitHub account was used to replace binaries on five ovhcloud-cli releases with a credential stealer for about 94 minutes. On three releases the checksums file was replaced too, so the hashes matched. StepSecurity traces the access to a Node backdoor from a malicious VS Code extension the developer installed a year earlier. A customer&#39;s CI caught the swap because it pinned hashes separately.</summary>
  </entry>

  <entry>
    <title>SubQuery @subql/common 5.8.3 published through a hijacked release workflow</title>
    <link href="https://intel.trapdex.com/campaigns/subql-common-release-workflow-hijack/"/>
    <id>https://intel.trapdex.com/campaigns/subql-common-release-workflow-hijack/</id>
    <published>2026-10-05T00:00:00Z</published>
    <updated>2026-10-05T00:00:00Z</updated>
    <summary>Using a maintainer&#39;s GitHub account, an attacker pushed short-lived branches that changed the release workflow. The workflow downloaded a tarball from an attacker domain and swapped it in for the real source before publishing through trusted-publisher OIDC. The resulting package carried a credential stealer with a remote shell, and loose version ranges pulled it into about 19 dependent packages.</summary>
  </entry>

  <entry>
    <title>Graphalgo: fake &#39;Veltrix Capital&#39; recruiters with malicious npm/PyPI graph libraries</title>
    <link href="https://intel.trapdex.com/campaigns/graphalgo-veltrix-capital-fake-recruiter/"/>
    <id>https://intel.trapdex.com/campaigns/graphalgo-veltrix-capital-fake-recruiter/</id>
    <published>2025-05-01T00:00:00Z</published>
    <updated>2026-09-24T00:00:00Z</updated>
    <summary>Recruiters for a fictional crypto trading firm gave JavaScript and Python developers interview tasks that pulled in lookalike graph and bignum libraries. The libraries were published clean and weaponized weeks later. The payload was a RAT that checks for MetaMask. In Sep 2026 Aikido tied Go modules and Terraform providers to the same cluster through shared infrastructure and a shared public key.</summary>
  </entry>

  <entry>
    <title>TraderTraitor Terraform &#39;candidate repos&#39; dropping FLATROOF and ROOFDECK</title>
    <link href="https://intel.trapdex.com/campaigns/tradertraitor-terraform-candidate-repos/"/>
    <id>https://intel.trapdex.com/campaigns/tradertraitor-terraform-candidate-repos/</id>
    <published>2026-09-19T00:00:00Z</published>
    <updated>2026-09-23T00:00:00Z</updated>
    <summary>DevOps candidates received interview projects whose .terraform.lock.hcl pointed a custom provider at typosquatted HashiCorp registry domains. Running terraform init fetched ARM64 Rust backdoors: FLATROOF for data theft and ROOFDECK for reconnaissance and lateral movement. SentinelOne documented a victim at an IT-services firm with no crypto ties.</summary>
  </entry>

  <entry>
    <title>Joint seven-agency advisory on WaterPlum / Contagious Interview (Dec 2025 - Jul 2026)</title>
    <link href="https://intel.trapdex.com/campaigns/waterplum-joint-advisory-2026/"/>
    <id>https://intel.trapdex.com/campaigns/waterplum-joint-advisory-2026/</id>
    <published>2025-12-01T00:00:00Z</published>
    <updated>2026-09-18T00:00:00Z</updated>
    <summary>Agencies from Japan, the US, Australia and Germany jointly reported that WaterPlum infected more than 30,000 devices in over 100 countries between Dec 2025 and Jul 2026 and took about USD 10.7M from more than 7,000 wallets. Lures impersonated AI, crypto and NFT firms, sometimes with face-swapped video calls. The malware set included BeaverTail, InvisibleFerret, OtterCookie, OtterCandy and StoatWaffle, some delivered through VS Code project autorun.</summary>
  </entry>

  <entry>
    <title>Mirage Kitten fake coding challenges delivering NodeRabbit and PollCat</title>
    <link href="https://intel.trapdex.com/campaigns/mirage-kitten-noderabbit-pollcat-challenges/"/>
    <id>https://intel.trapdex.com/campaigns/mirage-kitten-noderabbit-pollcat-challenges/</id>
    <published>2026-05-01T00:00:00Z</published>
    <updated>2026-09-01T00:00:00Z</updated>
    <summary>A non-DPRK case. Fake recruiters sent developers and fintech and aviation specialists in the Middle East and Africa coding-challenge archives hosted on S3. Trojanized packages bundled in node_modules launched NodeRabbit or PollCat, both cross-platform JavaScript RATs. The READMEs set a three-hour limit and banned AI assistants, likely so candidates would not review the code.</summary>
  </entry>

  <entry>
    <title>&#39;Superior&#39;: bought-and-trojanized extensions delivering a wallet drainer, credential theft and ClickFix</title>
    <link href="https://intel.trapdex.com/campaigns/superior-acquired-extensions-wallet-drainer/"/>
    <id>https://intel.trapdex.com/campaigns/superior-acquired-extensions-wallet-drainer/</id>
    <published>2024-01-01T00:00:00Z</published>
    <updated>2026-08-30T00:00:00Z</updated>
    <summary>Socket found 18 Chrome extensions and one Edge extension sharing a modular malware framework. At least five had been bought from their original developers and turned malicious through automatic updates, including a right-click and OCR tool with about 80k users. The modules hijack wallet buttons, show fake hardware-wallet &#39;restore&#39; pages, log credentials and history, and inject ClickFix fake-update prompts.</summary>
  </entry>

  <entry>
    <title>TerminalFix: fake CAPTCHA tells visitors to paste a command into Windows Terminal, leading to a reverse tunnel into the network</title>
    <link href="https://intel.trapdex.com/campaigns/terminalfix-fake-captcha-reverse-tunnel/"/>
    <id>https://intel.trapdex.com/campaigns/terminalfix-fake-captcha-reverse-tunnel/</id>
    <published>2026-08-28T00:00:00Z</published>
    <updated>2026-08-28T00:00:00Z</updated>
    <summary>A fake Cloudflare Turnstile check on compromised websites told visitors to paste a command into Windows Terminal or PowerShell. The command unpacked a ZIP, sideloaded a malicious DLL through a genuine Windows binary and pulled a payload hidden in PNG pixels. The end result is a Python reverse tunnel over port 443 that relays arbitrary TCP connections, which Microsoft says turns the PC into a stepping stone into the network.</summary>
  </entry>

  <entry>
    <title>Fake Breezy HR assessment with a &#39;camera fix&#39; ClickFix command (PylangGhost / GolangGhost)</title>
    <link href="https://intel.trapdex.com/campaigns/clickfake-breezyhr-camera-fix/"/>
    <id>https://intel.trapdex.com/campaigns/clickfake-breezyhr-camera-fix/</id>
    <published>2026-07-21T00:00:00Z</published>
    <updated>2026-07-28T00:00:00Z</updated>
    <summary>Crypto and Web3 staff, including non-developers, were sent to a typosquatted Breezy HR assessment site. At the video step the operator triggers a fake camera error. The suggested &#39;fix&#39; copies a different command than the one shown on screen, and that command installs a Python RAT on Windows or a Go RAT on macOS.</summary>
  </entry>

  <entry>
    <title>REF9403: fake contractor in a community Slack sends trojanized Next.js shop &#39;test challenges&#39;</title>
    <link href="https://intel.trapdex.com/campaigns/ref9403-community-slack-ecommerce-test/"/>
    <id>https://intel.trapdex.com/campaigns/ref9403-community-slack-ecommerce-test/</id>
    <published>2026-05-26T00:00:00Z</published>
    <updated>2026-07-18T00:00:00Z</updated>
    <summary>A persona advertising e-commerce contract work in a public community Slack moved candidates to DMs and sent backdoored copies of a public Next.js shop template. The code runs on npm run dev or npm start and delivers a four-stage OtterCookie chain: browser and wallet stealer, file stealer, Socket.IO RAT, and clipboard stealer.</summary>
  </entry>

  <entry>
    <title>Fake Zoom, Meet and Teams &quot;update to join&quot; pages install a legitimate remote monitoring tool configured for the attacker</title>
    <link href="https://intel.trapdex.com/campaigns/fake-meeting-update-signed-rmm/"/>
    <id>https://intel.trapdex.com/campaigns/fake-meeting-update-signed-rmm/</id>
    <published>2026-02-01T00:00:00Z</published>
    <updated>2026-06-30T00:00:00Z</updated>
    <summary>Meeting invitations lead to lookalike Zoom, Google Meet or Teams pages, sometimes with a fake waiting room or participant list, that say the client is out of date. The &#39;update&#39; is a genuine, signed remote monitoring or employee monitoring product set up for the attacker&#39;s account, which gives them silent access. Netskope, Malwarebytes and the Swiss federal cyber office reported waves from February to June 2026.</summary>
  </entry>

  <entry>
    <title>Mastra AI-agent framework: about 144 npm packages poisoned after a contributor was lured via LinkedIn</title>
    <link href="https://intel.trapdex.com/campaigns/mastra-contributor-linkedin-lure-easy-day-js/"/>
    <id>https://intel.trapdex.com/campaigns/mastra-contributor-linkedin-lure-easy-day-js/</id>
    <published>2026-06-16T00:00:00Z</published>
    <updated>2026-06-19T00:00:00Z</updated>
    <summary>A Mastra contributor was approached from a compromised LinkedIn account and clicked a malicious link during a call, which compromised their machine. The attacker published a clean dayjs lookalike, weaponized it a day later, and within about 30 minutes republished roughly 144 packages in the @mastra scope with that dependency added. Its postinstall drops a crypto-focused RAT with persistence on all three desktop platforms.</summary>
  </entry>

  <entry>
    <title>Malicious Nx Console 18.95.0 published to the VS Code Marketplace with stolen credentials</title>
    <link href="https://intel.trapdex.com/campaigns/nx-console-18-95-0-compromise/"/>
    <id>https://intel.trapdex.com/campaigns/nx-console-18-95-0-compromise/</id>
    <published>2026-05-18T00:00:00Z</published>
    <updated>2026-05-28T00:00:00Z</updated>
    <summary>After a developer&#39;s GitHub credentials leaked, a malicious nrwl.angular-console 18.95.0 was published outside the project&#39;s normal CI and stayed online for minutes. It ran an obfuscated stealer targeting GitHub, npm, AWS, Vault, Kubernetes and 1Password secrets, and on macOS it added a Python backdoor. GitHub said the build reached an employee laptop, which led to theft of internal repositories.</summary>
  </entry>

  <entry>
    <title>node-ipc dormant maintainer account taken over through an expired email domain</title>
    <link href="https://intel.trapdex.com/campaigns/node-ipc-expired-maintainer-domain-takeover/"/>
    <id>https://intel.trapdex.com/campaigns/node-ipc-expired-maintainer-domain-takeover/</id>
    <published>2026-05-07T00:00:00Z</published>
    <updated>2026-05-14T00:00:00Z</updated>
    <summary>A dormant maintainer&#39;s email domain lapsed and was re-registered on 7 May 2026, which let the new owner reset the npm password and gain publish rights. Three node-ipc versions then shipped an obfuscated stealer in the CommonJS entry point that exfiltrates developer and cloud credentials over HTTPS and DNS TXT queries.</summary>
  </entry>

  <entry>
    <title>GlassWorm &#39;sleeper&#39; wave: 73 cloned Open VSX extensions activated by later updates</title>
    <link href="https://intel.trapdex.com/campaigns/glassworm-73-openvsx-sleepers/"/>
    <id>https://intel.trapdex.com/campaigns/glassworm-73-openvsx-sleepers/</id>
    <published>2026-04-25T00:00:00Z</published>
    <updated>2026-04-29T00:00:00Z</updated>
    <summary>Socket tracked 73 copies of legitimate extensions that were published clean and later made malicious in batches through updates. Six were activated on 29 Apr 2026 with bundled native binaries, JavaScript that fetches payloads from GitHub, or extensionPack-based delivery. The rest were still dormant when Socket published.</summary>
  </entry>

  <entry>
    <title>@bitwarden/cli 2026.4.0 published through a poisoned Checkmarx GitHub Action</title>
    <link href="https://intel.trapdex.com/campaigns/bitwarden-cli-poisoned-checkmarx-action/"/>
    <id>https://intel.trapdex.com/campaigns/bitwarden-cli-poisoned-checkmarx-action/</id>
    <published>2026-04-22T00:00:00Z</published>
    <updated>2026-04-23T00:00:00Z</updated>
    <summary>A compromised Checkmarx GitHub Action in Bitwarden&#39;s release pipeline let the attacker publish a malicious @bitwarden/cli 2026.4.0 through the trusted OIDC path, and it was live for about 1.5 hours. A preinstall-triggered payload stole tokens, SSH keys, cloud credentials and AI-tool configs, and sent them to a Checkmarx-themed domain and to GitHub repos labelled &#39;Shai-Hulud: The Third Coming&#39;.</summary>
  </entry>

  <entry>
    <title>UNC6692: email bombing, then a fake helpdesk on Teams sends a &quot;spam patch&quot; that installs the SNOW malware suite</title>
    <link href="https://intel.trapdex.com/campaigns/unc6692-teams-helpdesk-snow/"/>
    <id>https://intel.trapdex.com/campaigns/unc6692-teams-helpdesk-snow/</id>
    <published>2025-12-01T00:00:00Z</published>
    <updated>2026-04-23T00:00:00Z</updated>
    <summary>After an email bombing campaign, UNC6692 messaged employees on Microsoft Teams as helpdesk staff offering help with the flood, and sent a link to a fake &#39;Mailbox Repair and Sync Utility&#39; page. The &#39;local patch&#39; it served ran an AutoHotkey script that installed a malicious Chromium extension (SNOWBELT), a tunneler (SNOWGLAZE) and a bind shell (SNOWBASIN). The intrusion moved on to credential dumping and domain controller compromise.</summary>
  </entry>

  <entry>
    <title>Void Dokkaebi interview repos that spread into the victim&#39;s own repositories</title>
    <link href="https://intel.trapdex.com/campaigns/void-dokkaebi-self-propagating-interview-repos/"/>
    <id>https://intel.trapdex.com/campaigns/void-dokkaebi-self-propagating-interview-repos/</id>
    <published>2026-04-21T00:00:00Z</published>
    <updated>2026-04-21T00:00:00Z</updated>
    <summary>Interview repos run a folder-open task that infects the developer, then push the loader into the victim&#39;s own repos and hide it by rewriting git history. Trend Micro counted more than 750 infected repos, some belonging to real organizations. The payload is the DEV#POPPER Node RAT, with C2 configuration fetched from the Tron, Aptos and BSC blockchains.</summary>
  </entry>

  <entry>
    <title>108 Chrome extensions on shared C2 stealing Google identity and Telegram Web sessions</title>
    <link href="https://intel.trapdex.com/campaigns/chrome-108-extensions-shared-c2-session-theft/"/>
    <id>https://intel.trapdex.com/campaigns/chrome-108-extensions-shared-c2-session-theft/</id>
    <published>2022-04-30T00:00:00Z</published>
    <updated>2026-04-14T00:00:00Z</updated>
    <summary>One shared C2 served 108 extensions (games, Telegram clients and translators) spread over five publisher names. Some collected Google account identity through OAuth2, some carried a backdoor that opens arbitrary URLs at browser start, and others sent Telegram Web sessions out every 15 seconds or stripped security headers to inject gambling overlays. About 20k installs.</summary>
  </entry>

  <entry>
    <title>Fake &#39;leaked Claude Code&#39; and AI-tool GitHub releases delivering Vidar and GhostSocks</title>
    <link href="https://intel.trapdex.com/campaigns/claude-code-leak-lure-github-releases/"/>
    <id>https://intel.trapdex.com/campaigns/claude-code-leak-lure-github-releases/</id>
    <published>2026-04-01T00:00:00Z</published>
    <updated>2026-04-03T00:00:00Z</updated>
    <summary>A day after the accidental Claude Code source-map exposure on 31 Mar 2026, attackers set up SEO-boosted GitHub repos offering &#39;leaked&#39; builds with unlocked features. Downloads delivered a Rust dropper, Vidar stealer and the GhostSocks proxy. Trend Micro says the same operation impersonates more than 25 brands, many of them AI tools.</summary>
  </entry>

  <entry>
    <title>axios npm maintainer lured with a fake company and Teams call; plain-crypto-js RAT dropper</title>
    <link href="https://intel.trapdex.com/campaigns/axios-maintainer-fake-meeting-plain-crypto-js/"/>
    <id>https://intel.trapdex.com/campaigns/axios-maintainer-fake-meeting-plain-crypto-js/</id>
    <published>2026-03-01T00:00:00Z</published>
    <updated>2026-03-31T00:00:00Z</updated>
    <summary>For about two weeks the lead axios maintainer was worked with a cloned company, a fake Slack workspace and a fake Teams meeting, which pushed a fake &#39;update&#39; that was a RAT. The attacker then published axios 1.14.1 and 0.30.4, both adding a pre-staged typosquat dependency whose postinstall installs a cross-platform backdoor. npm removed the versions about three hours after publication.</summary>
  </entry>

  <entry>
    <title>Dormant IoliteLabs Solidity extensions made malicious after eight years</title>
    <link href="https://intel.trapdex.com/campaigns/iolitelabs-solidity-dormant-takeover/"/>
    <id>https://intel.trapdex.com/campaigns/iolitelabs-solidity-dormant-takeover/</id>
    <published>2026-03-25T00:00:00Z</published>
    <updated>2026-03-28T00:00:00Z</updated>
    <summary>Three Solidity extensions first published in 2018 (about 27.5k installs) received a 0.1.8 update. The update hid a dropper inside a modified bundled compression library. On Windows it delivered a fake Chrome-updater MSI with a hook DLL, and on macOS it delivered Node-based binaries. The targets were Solidity and crypto developers.</summary>
  </entry>

  <entry>
    <title>TeamPCP backdoors the LiteLLM AI gateway SDK on PyPI (1.82.7 and 1.82.8)</title>
    <link href="https://intel.trapdex.com/campaigns/teampcp-litellm-1-82-pypi-compromise/"/>
    <id>https://intel.trapdex.com/campaigns/teampcp-litellm-1-82-pypi-compromise/</id>
    <published>2026-03-24T00:00:00Z</published>
    <updated>2026-03-25T00:00:00Z</updated>
    <summary>With a PyPI token stolen through the earlier Trivy compromise, the attacker published two LiteLLM versions carrying a .pth file that runs a credential stealer whenever Python starts. It targeted cloud, SSH, Kubernetes, LLM-provider and wallet secrets. PyPI quarantined the package after about three hours.</summary>
  </entry>

  <entry>
    <title>EvilTokens: phishing kit gets Microsoft 365 users to enter a device code on the real Microsoft sign-in page, handing over their session</title>
    <link href="https://intel.trapdex.com/campaigns/eviltokens-device-code-phishing/"/>
    <id>https://intel.trapdex.com/campaigns/eviltokens-device-code-phishing/</id>
    <published>2026-02-19T00:00:00Z</published>
    <updated>2026-03-23T00:00:00Z</updated>
    <summary>EvilTokens, a phishing-as-a-service kit advertised on Telegram from 16 Feb 2026, sent tailored lures (construction bids, DocuSign, Microsoft Forms, shared files) that ended with the victim typing an attacker-supplied code into the genuine Microsoft device login page. That signs the attacker&#39;s session in as the victim. Huntress counted more than 340 affected organizations in the US, Canada, Australia, New Zealand and Germany, with tokens replayed from Railway.com infrastructure.</summary>
  </entry>

  <entry>
    <title>Trivy release and trivy-action / setup-trivy tag hijack</title>
    <link href="https://intel.trapdex.com/campaigns/trivy-teampcp-action-tag-hijack/"/>
    <id>https://intel.trapdex.com/campaigns/trivy-teampcp-action-tag-hijack/</id>
    <published>2026-03-19T00:00:00Z</published>
    <updated>2026-03-23T00:00:00Z</updated>
    <summary>After an earlier Trivy incident disclosed on 1 Mar, credential rotation was incomplete and the attacker kept access. On 19 Mar they published a malicious Trivy 0.69.4 release and force-moved most trivy-action tags plus the setup-trivy tags to credential-stealing code, then added malicious Docker images on 22 Mar. A Checkmarx GitHub Action was hit in parallel, and stolen tokens led to the LiteLLM compromise.</summary>
  </entry>

  <entry>
    <title>InstallFix: malvertised clones of the Claude Code install page delivering Amatera Stealer</title>
    <link href="https://intel.trapdex.com/campaigns/installfix-fake-claude-code-install-pages/"/>
    <id>https://intel.trapdex.com/campaigns/installfix-fake-claude-code-install-pages/</id>
    <published>2026-03-06T00:00:00Z</published>
    <updated>2026-03-16T00:00:00Z</updated>
    <summary>Push Security reported sponsored search results leading to near-identical copies of the Claude Code install page, with the one-line install command swapped for one that fetches an infostealer. The main payload observed was Amatera Stealer. Similar clones targeted other developer tools.</summary>
  </entry>

  <entry>
    <title>GlassWorm March 2026 wave: 72 Open VSX extensions using extensionPack delivery, plus a trojanized MCP server</title>
    <link href="https://intel.trapdex.com/campaigns/glassworm-72-openvsx-transitive-and-mcp/"/>
    <id>https://intel.trapdex.com/campaigns/glassworm-72-openvsx-transitive-and-mcp/</id>
    <published>2026-03-12T00:00:00Z</published>
    <updated>2026-03-14T00:00:00Z</updated>
    <summary>At least 72 more Open VSX extensions were tied to GlassWorm. Most declared a separate malicious extension in extensionPack or extensionDependencies, so a harmless-looking listing could pull in the loader after a later manifest change. The same wave published a lookalike of the WaterCrawl MCP server to npm with an invisible-Unicode payload, and touched more than 150 GitHub repos.</summary>
  </entry>

  <entry>
    <title>Fake OpenClaw installer repos, recommended by AI search, delivering Vidar, GhostSocks and AMOS</title>
    <link href="https://intel.trapdex.com/campaigns/fake-openclaw-installer-repos/"/>
    <id>https://intel.trapdex.com/campaigns/fake-openclaw-installer-repos/</id>
    <published>2026-02-02T00:00:00Z</published>
    <updated>2026-03-05T00:00:00Z</updated>
    <summary>Huntress found GitHub repos posing as OpenClaw installers, padded with copied legitimate code, that a search engine&#39;s AI answers recommended. They delivered Rust loaders, Vidar and the GhostSocks proxy on Windows, and Atomic Stealer on macOS. GitHub took them down around 10 Feb.</summary>
  </entry>

  <entry>
    <title>hackerbot-claw: self-described AI agent exploiting GitHub Actions and poisoning CLAUDE.md</title>
    <link href="https://intel.trapdex.com/campaigns/hackerbot-claw-ai-agent-actions-exploitation/"/>
    <id>https://intel.trapdex.com/campaigns/hackerbot-claw-ai-agent-actions-exploitation/</id>
    <published>2026-02-20T00:00:00Z</published>
    <updated>2026-03-02T00:00:00Z</updated>
    <summary>A GitHub account describing itself as an autonomous agent attacked CI workflows in several large open-source projects using pwn-request, branch-name, filename and script injection. It got code execution in several repos and stole a write-scoped token from one. In another repo it replaced CLAUDE.md with injected instructions aimed at an AI review workflow, which the model flagged and refused.</summary>
  </entry>

  <entry>
    <title>Cline CLI 2.3.0 npm release hijacked after prompt injection in an AI triage workflow</title>
    <link href="https://intel.trapdex.com/campaigns/cline-cli-2-3-0-prompt-injection-token-theft/"/>
    <id>https://intel.trapdex.com/campaigns/cline-cli-2-3-0-prompt-injection-token-theft/</id>
    <published>2026-02-17T00:00:00Z</published>
    <updated>2026-02-24T00:00:00Z</updated>
    <summary>A stolen npm token was used to publish cline@2.3.0, whose only change was a postinstall script that globally installs the OpenClaw agent. Cline&#39;s post-mortem traces the token theft to a prompt injection in a GitHub issue title, which steered an LLM issue-triage workflow, combined with GitHub Actions cache poisoning. About 4,000 downloads happened in the roughly eight-hour window.</summary>
  </entry>

  <entry>
    <title>Recruiting-themed Next.js repos with three backdoor execution paths</title>
    <link href="https://intel.trapdex.com/campaigns/nextjs-interview-repos-three-execution-paths/"/>
    <id>https://intel.trapdex.com/campaigns/nextjs-interview-repos-three-execution-paths/</id>
    <published>2026-02-24T00:00:00Z</published>
    <updated>2026-02-24T00:00:00Z</updated>
    <summary>Microsoft found Next.js assessment repos that launch a JavaScript backdoor in one of three ways: a VS Code folder-open task, a trojanized jquery.min.js at build time, or a backend module that decodes a C2 address from .env at server start. The first stage profiles the host, and the second stage provides file browsing and staged exfiltration.</summary>
  </entry>

  <entry>
    <title>SANDWORM_MODE: npm worm that plants rogue MCP servers in AI assistant configs</title>
    <link href="https://intel.trapdex.com/campaigns/sandworm-mode-npm-worm-mcp-injection/"/>
    <id>https://intel.trapdex.com/campaigns/sandworm-mode-npm-worm-mcp-injection/</id>
    <published>2026-02-20T00:00:00Z</published>
    <updated>2026-02-20T00:00:00Z</updated>
    <summary>Socket found a Shai-Hulud-style worm spread through at least 19 typosquatted npm packages, some named after AI coding tools. One module writes a rogue MCP server into local AI-assistant configs. The server&#39;s tool descriptions carry hidden instructions telling the assistant to collect SSH keys, cloud credentials, npm tokens and env secrets. The worm spreads using stolen npm and GitHub tokens.</summary>
  </entry>

  <entry>
    <title>AiFrame: 30+ fake AI-assistant extensions run from a remote-controlled iframe</title>
    <link href="https://intel.trapdex.com/campaigns/aiframe-fake-ai-assistant-iframes/"/>
    <id>https://intel.trapdex.com/campaigns/aiframe-fake-ai-assistant-iframes/</id>
    <published>2026-02-06T00:00:00Z</published>
    <updated>2026-02-13T00:00:00Z</updated>
    <summary>More than 30 Chrome extensions posing as Gemini, ChatGPT and Claude sidebars, translators and Gmail helpers loaded their whole UI as an iframe from operator servers, so the operator could change behavior at any time. They collected page content and browsing data, and 15 of them read Gmail threads and drafts. About 260k installs; removed extensions reappeared under new IDs.</summary>
  </entry>

  <entry>
    <title>CL Suite: Meta Business Manager helper that steals exports and 2FA seeds</title>
    <link href="https://intel.trapdex.com/campaigns/cl-suite-meta-business-totp-stealer/"/>
    <id>https://intel.trapdex.com/campaigns/cl-suite-meta-business-totp-stealer/</id>
    <published>2025-03-01T00:00:00Z</published>
    <updated>2026-02-13T00:00:00Z</updated>
    <summary>An extension sold to Meta ads operators as a Business Manager scraper, popup remover and built-in 2FA generator sent TOTP seeds and current codes, Business Manager people exports and analytics to an attacker backend, with optional forwarding to Telegram. This defeats 2FA on the accounts involved. Its privacy policy falsely said all data stayed local.</summary>
  </entry>

  <entry>
    <title>ClawHavoc: malicious OpenClaw skills on ClawHub delivering Atomic Stealer</title>
    <link href="https://intel.trapdex.com/campaigns/clawhavoc-clawhub-amos-skills/"/>
    <id>https://intel.trapdex.com/campaigns/clawhavoc-clawhub-amos-skills/</id>
    <published>2026-02-02T00:00:00Z</published>
    <updated>2026-02-02T00:00:00Z</updated>
    <summary>Koi audited the 2,857 skills on ClawHub, the OpenClaw skill registry, and found 341 malicious ones. 335 belonged to one coordinated operation. The skill instructions listed fake &#39;prerequisite&#39; install steps that got the user, or the agent itself, to run commands that installed Atomic Stealer on macOS or a trojan archive on Windows.</summary>
  </entry>

  <entry>
    <title>Open VSX publisher &#39;oorzc&#39; hijacked to push GlassWorm loader updates</title>
    <link href="https://intel.trapdex.com/campaigns/openvsx-oorzc-publisher-hijack-glassworm/"/>
    <id>https://intel.trapdex.com/campaigns/openvsx-oorzc-publisher-hijack-glassworm/</id>
    <published>2026-01-30T00:00:00Z</published>
    <updated>2026-02-02T00:00:00Z</updated>
    <summary>Malicious updates landed in four established extensions from one real publisher (over 22k downloads), most likely through a leaked publish token. It was the first time GlassWorm used a genuine developer account rather than fake publishers. Open VSX revoked the tokens and removed the bad versions.</summary>
  </entry>

  <entry>
    <title>BigBlack VS Code extensions delivering Evelyn Stealer</title>
    <link href="https://intel.trapdex.com/campaigns/bigblack-extensions-evelyn-stealer/"/>
    <id>https://intel.trapdex.com/campaigns/bigblack-extensions-evelyn-stealer/</id>
    <published>2025-12-08T00:00:00Z</published>
    <updated>2026-01-28T00:00:00Z</updated>
    <summary>Koi first flagged three BigBlack extensions that dropped a downloader DLL. Trend Micro later traced the full chain: a fake Lightshot.dll, an injector using process hollowing, and Evelyn Stealer. The stealer takes browser credentials, wallets, clipboard data, Wi-Fi keys and screenshots, and uploads them over FTP.</summary>
  </entry>

  <entry>
    <title>&#39;Fake Font&#39; assignment repos hiding JavaScript loaders as .woff2 files</title>
    <link href="https://intel.trapdex.com/campaigns/contagious-code-fake-font-woff2/"/>
    <id>https://intel.trapdex.com/campaigns/contagious-code-fake-font-woff2/</id>
    <published>2026-01-16T00:00:00Z</published>
    <updated>2026-01-28T00:00:00Z</updated>
    <summary>In this variant, fintech and crypto recruiters sent about 17-19 assignment repos. Each repo&#39;s tasks.json executed JavaScript disguised as a web-font file. The chain ran from BeaverTail to an InvisibleFerret loader and backdoor, which steal wallets and credentials, and was staged on a domain impersonating Alchemy.</summary>
  </entry>

  <entry>
    <title>Fake &#39;ClawdBot Agent&#39; AI extension installing a ScreenConnect RAT</title>
    <link href="https://intel.trapdex.com/campaigns/fake-clawdbot-agent-screenconnect/"/>
    <id>https://intel.trapdex.com/campaigns/fake-clawdbot-agent-screenconnect/</id>
    <published>2026-01-27T00:00:00Z</published>
    <updated>2026-01-28T00:00:00Z</updated>
    <summary>A working multi-provider AI assistant posing as Clawdbot/Moltbot, which has no official VS Code extension, ran at every VS Code start. It fetched a remote config, installed a ScreenConnect client disguised as Code.exe and pointed at an attacker relay, and carried a Rust DLL fallback that pulled from Dropbox. Microsoft removed it after Aikido reported it, at about 21 installs.</summary>
  </entry>

  <entry>
    <title>MaliciousCorgi: working AI-assistant extensions that send every opened file to China-hosted servers</title>
    <link href="https://intel.trapdex.com/campaigns/maliciouscorgi-ai-assistant-source-exfil/"/>
    <id>https://intel.trapdex.com/campaigns/maliciouscorgi-ai-assistant-source-exfil/</id>
    <published>2026-01-23T00:00:00Z</published>
    <updated>2026-01-26T00:00:00Z</updated>
    <summary>Two working ChatGPT-style coding assistants with about 1.5M combined installs uploaded the full contents of every opened file, along with edits, to remote servers. They could also run server-triggered harvesting of up to 50 workspace files, and they loaded analytics SDKs in hidden iframes to fingerprint users.</summary>
  </entry>

  <entry>
    <title>Assessment repos that autorun through VS Code tasks.json folderOpen</title>
    <link href="https://intel.trapdex.com/campaigns/contagious-interview-vscode-tasks-folderopen/"/>
    <id>https://intel.trapdex.com/campaigns/contagious-interview-vscode-tasks-folderopen/</id>
    <published>2025-12-01T00:00:00Z</published>
    <updated>2026-01-20T00:00:00Z</updated>
    <summary>Fake CTOs and recruiters shared Git repos presented as hiring assessments. Each contained a hidden .vscode/tasks.json set to run on folder open, so trusting the workspace in VS Code was enough to pipe a remote script into Node. The script installed a JavaScript backdoor that beacons every few seconds.</summary>
  </entry>

  <entry>
    <title>Fake Workday, NetSuite and SuccessFactors tool extensions hijacking enterprise sessions</title>
    <link href="https://intel.trapdex.com/campaigns/hr-erp-session-hijack-extensions/"/>
    <id>https://intel.trapdex.com/campaigns/hr-erp-session-hijack-extensions/</id>
    <published>2021-08-18T00:00:00Z</published>
    <updated>2026-01-16T00:00:00Z</updated>
    <summary>Five linked extensions sold as &#39;premium tools&#39; for HR and ERP platforms. They sent session cookies to remote servers, injected stolen cookies to take over accounts, and blocked security and admin pages to slow incident response. They had about 2,300 installs across two publisher names.</summary>
  </entry>

  <entry>
    <title>GhostPoster: JavaScript loader hidden in extension PNG icons</title>
    <link href="https://intel.trapdex.com/campaigns/ghostposter-png-steganography-extensions/"/>
    <id>https://intel.trapdex.com/campaigns/ghostposter-png-steganography-extensions/</id>
    <published>2020-01-01T00:00:00Z</published>
    <updated>2026-01-15T00:00:00Z</updated>
    <summary>Koi reported 17 Firefox add-ons, including VPNs, translators, weather and ad blockers, whose logo PNG hid a loader. The loader waits about 48 hours and fetches a payload only about 10% of the time. The payload hijacks affiliate links, injects tracking, strips security headers and bypasses CAPTCHAs. LayerX then linked about 17 more extensions across Chrome, Edge and Firefox, with about 840k total installs.</summary>
  </entry>

  <entry>
    <title>Cloned AI-sidebar extensions (one with a Featured badge) exfiltrating ChatGPT and DeepSeek chats</title>
    <link href="https://intel.trapdex.com/campaigns/featured-ai-sidebar-clones-chat-exfil/"/>
    <id>https://intel.trapdex.com/campaigns/featured-ai-sidebar-clones-chat-exfil/</id>
    <published>2025-12-29T00:00:00Z</published>
    <updated>2026-01-07T00:00:00Z</updated>
    <summary>Two copies of a legitimate AI-sidebar extension, with about 900k installs between them and one carrying Google&#39;s Featured badge, asked for consent to &#39;anonymous analytics&#39;. Every 30 minutes they uploaded full ChatGPT and DeepSeek conversations and all tab URLs, including any session tokens in URL parameters. Parts of the operator&#39;s infrastructure were hosted on an AI app-builder platform.</summary>
  </entry>

  <entry>
    <title>Zoom Stealer: meeting-helper extensions harvesting corporate meeting data</title>
    <link href="https://intel.trapdex.com/campaigns/darkspectre-zoom-stealer/"/>
    <id>https://intel.trapdex.com/campaigns/darkspectre-zoom-stealer/</id>
    <published>2025-12-31T00:00:00Z</published>
    <updated>2025-12-31T00:00:00Z</updated>
    <summary>Eighteen extensions sold as video-meeting helpers requested access to more than 28 conferencing platforms. They collected meeting URLs with embedded passcodes, participant lists, speaker bios and schedules from about 2.2M users. Koi described it as corporate-espionage infrastructure and said further extensions were dormant and ready to be activated.</summary>
  </entry>

  <entry>
    <title>Trust Wallet Chrome extension 2.68 published with a leaked store API key</title>
    <link href="https://intel.trapdex.com/campaigns/trust-wallet-extension-2-68-compromise/"/>
    <id>https://intel.trapdex.com/campaigns/trust-wallet-extension-2-68-compromise/</id>
    <published>2025-12-24T00:00:00Z</published>
    <updated>2025-12-30T00:00:00Z</updated>
    <summary>Developer secrets exposed in the Shai-Hulud 2.0 npm worm included a Chrome Web Store API key. Attackers used it to publish version 2.68.0 of the official Trust Wallet extension with code that sent seed phrases to a lookalike domain. Trust Wallet reports about 2,520 wallets drained of about USD 8.5M between 24 and 26 Dec. Version 2.69 removed the code, and a follow-up phishing site targeted users looking for a fix.</summary>
  </entry>

  <entry>
    <title>GlassWorm moves to macOS with encrypted payloads and wallet-app replacement code</title>
    <link href="https://intel.trapdex.com/campaigns/glassworm-macos-wave/"/>
    <id>https://intel.trapdex.com/campaigns/glassworm-macos-wave/</id>
    <published>2025-12-29T00:00:00Z</published>
    <updated>2025-12-29T00:00:00Z</updated>
    <summary>Koi reported three Open VSX extensions (about 50k downloads) carrying AES-encrypted JavaScript that waits 15 minutes before running on macOS. The payload persists through a LaunchAgent, reads the Keychain, and contains code to replace hardware-wallet apps with trojanized copies. Shared IPs and a shared Solana address tie it to earlier GlassWorm waves.</summary>
  </entry>

  <entry>
    <title>Urban VPN Proxy and sister extensions capturing AI chatbot conversations</title>
    <link href="https://intel.trapdex.com/campaigns/urban-vpn-ai-chat-harvesting/"/>
    <id>https://intel.trapdex.com/campaigns/urban-vpn-ai-chat-harvesting/</id>
    <published>2025-07-09T00:00:00Z</published>
    <updated>2025-12-18T00:00:00Z</updated>
    <summary>According to The Register and Malwarebytes, reporting on research by Koi Security, version 5.5.0 of the free Urban VPN Proxy extension and related extensions from the same publisher added scripts that capture prompts and responses on AI chat sites, including ChatGPT, Claude, Gemini, Microsoft Copilot, Perplexity, DeepSeek, Grok and Meta AI. The Register reports that collection was on by default with no user setting, that the change arrived in a silent update in July 2025, and that the publisher&#39;s privacy policy says the data is sold for marketing. The extensions had more than 8 million installs.</summary>
  </entry>

  <entry>
    <title>19 theme extensions hiding a trojan in a modified bundled dependency and a fake PNG</title>
    <link href="https://intel.trapdex.com/campaigns/theme-remakes-fake-png-trojan/"/>
    <id>https://intel.trapdex.com/campaigns/theme-remakes-fake-png-trojan/</id>
    <published>2025-02-01T00:00:00Z</published>
    <updated>2025-12-10T00:00:00Z</updated>
    <summary>ReversingLabs found 19 extensions, mostly theme &#39;remakes&#39;, that shipped a tampered copy of a common npm module inside their bundled node_modules. The tampered code ran when VS Code started and decoded a dropper. The dropper unpacked a fake banner.png archive and launched a Rust trojan through cmstp.exe.</summary>
  </entry>

  <entry>
    <title>GlassWorm third wave: 24 impersonating extensions on VS Code Marketplace and Open VSX</title>
    <link href="https://intel.trapdex.com/campaigns/glassworm-wave3-24-impersonating-extensions/"/>
    <id>https://intel.trapdex.com/campaigns/glassworm-wave3-24-impersonating-extensions/</id>
    <published>2025-12-02T00:00:00Z</published>
    <updated>2025-12-02T00:00:00Z</updated>
    <summary>Secure Annex found 24 GlassWorm extensions copying popular tools such as Flutter, Vim, YAML, Tailwind, Svelte and Claude helpers. They were approved while clean and then updated with malicious code. The updates used invisible-Unicode obfuscation and bundled Rust native implants, the listings had inflated download counts, and C2 was resolved through Solana and Google Calendar dead drops.</summary>
  </entry>

  <entry>
    <title>ShadyPanda: years-long sleeper extensions turned into a backdoor and spyware (Clean Master, WeTab, Infinity New Tab)</title>
    <link href="https://intel.trapdex.com/campaigns/shadypanda-sleeper-extensions/"/>
    <id>https://intel.trapdex.com/campaigns/shadypanda-sleeper-extensions/</id>
    <published>2018-01-01T00:00:00Z</published>
    <updated>2025-12-02T00:00:00Z</updated>
    <summary>An operator published useful-looking new-tab, cleaner and wallpaper extensions, built up installs and trust badges over years, then pushed malicious updates. One family received an hourly remote-code backdoor (about 300k users). Another, with about 4M users, carried full browsing surveillance that sent data to servers in China. Google and Microsoft removed the listings, but existing installs stay until users remove them.</summary>
  </entry>

  <entry>
    <title>Contagious Interview npm &#39;test assignment&#39; wave delivering OtterCookie (197+ packages)</title>
    <link href="https://intel.trapdex.com/campaigns/contagious-interview-npm-test-assignments-ottercookie-v4/"/>
    <id>https://intel.trapdex.com/campaigns/contagious-interview-npm-test-assignments-ottercookie-v4/</id>
    <published>2025-10-10T00:00:00Z</published>
    <updated>2025-11-26T00:00:00Z</updated>
    <summary>Fake recruiters sent Web3 developers coding tests whose projects depended on lookalike npm packages. Socket counted 197 new malicious packages from 10 Oct 2025, with more than 31k downloads. The packages fetched a payload from GitHub through Vercel staging apps and installed an OtterCookie build that merges BeaverTail loader functions with stealer and RAT features.</summary>
  </entry>

  <entry>
    <title>Lookalike Microsoft login domain impersonates a Czech company&#39;s IT department to deliver a ClickFix RAT</title>
    <link href="https://intel.trapdex.com/campaigns/czech-it-dept-lookalike-clickfix-rat/"/>
    <id>https://intel.trapdex.com/campaigns/czech-it-dept-lookalike-clickfix-rat/</id>
    <published>2025-11-25T00:00:00Z</published>
    <updated>2025-11-25T00:00:00Z</updated>
    <summary>A Microsoft lookalike domain hosted a Czech-language &#39;verify you are human&#39; page styled as one company&#39;s IT department. It told staff to press Win+R, paste and run a command, which installed two UPX-packed PureBasic loaders named like Windows components. The loaders unpack batch scripts that act as a remote access trojan, taking commands from a small PHP panel.</summary>
  </entry>

  <entry>
    <title>Shai-Hulud 2.0 (&#39;Sha1-Hulud: The Second Coming&#39;) self-propagating npm worm</title>
    <link href="https://intel.trapdex.com/campaigns/shai-hulud-2-npm-worm/"/>
    <id>https://intel.trapdex.com/campaigns/shai-hulud-2-npm-worm/</id>
    <published>2025-11-21T00:00:00Z</published>
    <updated>2025-11-25T00:00:00Z</updated>
    <summary>A worm spread through npm using stolen maintainer tokens. Each infected package gained a preinstall hook that runs a Bun-based stealer, which posts developer, CI and cloud credentials to public GitHub repos and registers infected hosts as self-hosted runners. Datadog counted 796 backdoored packages with more than 20M weekly downloads.</summary>
  </entry>

  <entry>
    <title>&#39;Safery: Ethereum Wallet&#39;: fake wallet extension leaking seed phrases through Sui micro-transactions</title>
    <link href="https://intel.trapdex.com/campaigns/safery-fake-ethereum-wallet-sui-exfil/"/>
    <id>https://intel.trapdex.com/campaigns/safery-fake-ethereum-wallet-sui-exfil/</id>
    <published>2025-09-29T00:00:00Z</published>
    <updated>2025-11-13T00:00:00Z</updated>
    <summary>A fake Ethereum wallet that ranked near the top of store search encoded newly created or imported seed phrases into fake Sui addresses. It then sent tiny SUI amounts to those addresses from an attacker wallet, so the secret travels through public blockchain data rather than a normal C2 channel, which helps it avoid network-based detection.</summary>
  </entry>

  <entry>
    <title>Lookalike npm package impersonating Claude Code that intercepts API traffic</title>
    <link href="https://intel.trapdex.com/campaigns/chatgptclaude-club-fake-claude-code-npm/"/>
    <id>https://intel.trapdex.com/campaigns/chatgptclaude-club-fake-claude-code-npm/</id>
    <published>2025-10-27T00:00:00Z</published>
    <updated>2025-11-12T00:00:00Z</updated>
    <summary>Safety reported an npm package imitating the Claude Code CLI under a lookalike scope. It steals Claude credentials from environment variables and the user&#39;s config directory, then patches the real CLI with an interceptor that proxies API traffic. Prompts, conversations and account data go to a C2 server that can also send back commands.</summary>
  </entry>

  <entry>
    <title>Storm-1811: email bombing, then a fake IT helpdesk on Teams or by phone talks staff into a Quick Assist session</title>
    <link href="https://intel.trapdex.com/campaigns/storm-1811-email-bomb-teams-helpdesk-quick-assist/"/>
    <id>https://intel.trapdex.com/campaigns/storm-1811-email-bomb-teams-helpdesk-quick-assist/</id>
    <published>2024-04-01T00:00:00Z</published>
    <updated>2024-12-04T00:00:00Z</updated>
    <summary>The attacker signs an employee&#39;s address up to thousands of mailing lists, then contacts them as the IT helpdesk, by phone or from an external Microsoft Teams tenant, offering to fix the flood. The employee is talked into starting Quick Assist or installing a remote support tool. The attacker then drops loaders and remote tools, steals credentials and, in Microsoft&#39;s cases, deployed Black Basta ransomware.</summary>
  </entry>

</feed>
