VULNpocalypse
CriticalSupply chainActive

ARTEX Docker image backdoored, researcher says; CERT-In flags Crafty Falcon

A researcher reports a backdoored ARTEX Docker image giving unauthenticated root shells; remove any ARTEX install built from unvetted images.

Published
Exploitation
No confirmed use of the backdoor itself; CERT-In says ARTEX is among the tools used in active, ongoing attacks on BFSI entities

What happened

There are two related reports.

CERT-In advisory. CERT-In has issued advisory CIAD-2026-0058, rated Critical and dated November 04, 2026. According to the advisory, an AI-assisted threat actor that its CSIRT-Fin team has named “CRAFTY FALCON” is attacking Indian BFSI entities. CERT-In names the affected sectors as NBFCs, digital lending platforms, fintech entities, payment aggregators, payment service providers and cryptocurrency exchanges (CERT-In).

  • CERT-In calls the threat “active and ongoing”. It says entities that had already remediated after earlier restricted advisories have been deliberately re-targeted (CERT-In).
  • The advisory updates CIAD-2026-0046, dated October 09, 2026. In that earlier advisory, CERT-In describes stolen payment aggregator, gateway and partner-bank API credentials being used to call fund-transfer APIs directly, with funds credited to mule accounts across multiple banks (CERT-In).
  • CERT-In says its analysis identified several publicly available AI-driven offensive tools. These include the “ARTEX” family, “PentAGI”, the “Cairn” attack search engine and “CyberStrikeAI” (CERT-In).

CERT-In reports the following techniques (CERT-In):

  • Abuse of cloud management services such as AWS Systems Manager and Lambda.
  • Entry through internet-exposed UAT, staging and test environments, followed by movement into production.
  • Use of modified application packages to call APIs.
  • Micro-value test transactions to validate unauthorized fund flows before large-value transfers. CERT-In gives Rs. 1 as an example.
  • Exploitation of newly published file-upload and deserialization flaws, leading to web shells in PHP and Node.js applications.
  • Theft of cryptocurrency wallet keys and funds.

The public advisory names no CVE IDs. According to CERT-In, indicators of compromise were shared only in a restricted version.

Backdoored ARTEX copy. The Hacker News reports the following:

  • ARTEX’s developer took the repository down from GitHub after the tool was abused in real-world intrusions. Copies then appeared on GitHub.
  • An independent researcher using the alias Badbyte says the copied “trhhubsd” repository was starred 2,449 times and forked over 600 times.
  • According to the outlet, that copy changed one line in docker-compose.yml so that it pulls the tomicdd/artex image from Docker Hub. The outlet says Docker Hub shows 9,000 downloads of that image.
  • The researcher says the image adds a hidden verifyCaptcha() function to the login handler. If the CAPTCHA field starts with a secret marker, the rest of the field runs through sh -c as root before any credential check. The researcher describes this as an unauthenticated root shell.
  • The researcher says congdonglinux/artex is also backdoored and is identical to tomicdd/artex.
  • The researcher reports 62 internet-facing ARTEX consoles serving the backdoored build between November 2 and 4, 2026. The researcher also counted 1,645 hosts serving an ARTEX console as of November 4, 2026.

What is not confirmed:

  • The backdoor findings come from a single researcher, reported by one outlet. There is no independent verification in the sources.
  • Whether the backdoor has been used against anyone is not confirmed.
  • Who planted the backdoor is not confirmed.
  • The sources do not state any link between the backdoor and Crafty Falcon.

Am I affected?

  • You installed ARTEX from a GitHub copy. You may be affected if your docker-compose.yml references tomicdd/artex or congdonglinux/artex. The Hacker News lists the copies as trhhubsd/ARTEXX, mhtsec/ARTEX and RuoJi6/ARTEX (The Hacker News).
  • You use another “artex” image from Docker Hub. The outlet lists mianhuatang11/artex, autumn26/artex, andy1099zz/artex, dgddw/artex, trhhubsd/ARTEX and ylh440104/artex. It does not say whether these images are backdoored, so that is not confirmed (The Hacker News).
  • Your console is reachable from the internet. The researcher says the backdoor is triggered by an unauthenticated request to the login endpoint (The Hacker News).
  • You are an Indian BFSI entity. CERT-In advises all BFSI entities to treat the Crafty Falcon campaign as an immediate and continuing threat (CERT-In).

What to do now

  1. Find the image reference. Search docker-compose files and running containers for tomicdd/artex and congdonglinux/artex (The Hacker News).
  2. If either image ran, treat the host as compromised. CERT-In advises that, on detecting suspicious activity, you isolate the affected host, preserve forensic evidence and logs, and rotate all credentials (CERT-In). The researcher notes that a compromised ARTEX instance holds its targets’ findings, live credentials and captured sessions (The Hacker News).
  3. Control offensive tools on corporate systems. CERT-In recommends the following (CERT-In):
    • Prohibit unauthorized AI-driven offensive frameworks on corporate infrastructure.
    • Monitor for downloads of these tools.
    • Monitor for AI coding-agent processes running with permission checks disabled.
    • Monitor LLM API traffic coming from servers.
  4. Remove hardcoded secrets. CERT-In recommends the following (CERT-In):
    • Prohibit hardcoded tokens in source and build code.
    • Run secret scanning across repositories and CI/CD pipelines.
    • Use short-lived, least-privileged tokens.
    • Enforce branch protection, commit signing and code review on production branches.
  5. Patch internet-facing PHP and Node.js applications quickly. CERT-In recommends the following (CERT-In):
    • Prioritize file-upload and deserialization weaknesses.
    • Use WAF virtual patching as an interim control.
    • Make deployment directories read-only for the runtime user.
    • Add file-integrity monitoring on application directories.
  6. Separate non-production from production. CERT-In advises isolating UAT, staging and test environments from production, with no shared credentials, tokens or network paths (CERT-In).
  7. If you are a BFSI entity, follow the full advisories and report incidents. Apply the complete recommendations in CIAD-2026-0058 and CIAD-2026-0046. Report suspicious activity to CERT-In/CSIRT-Fin at incident@cert-in.org.in (CERT-In, CERT-In).
  8. If you build or audit AI agents, read the new guidelines. CERT-In has also published Artificial Intelligence Security Auditing Guidelines, dated November 04, 2026 (CERT-In).

Sources