Infiltrating the software supply chain is not a new attack method, but the way cybercriminals insinuate themselves and their malicious code into repositories continues to become more sophisticated. Although developers know that any open-source code should be reviewed and vetted, attackers now work to circumvent that practice.
In a recent campaign targeting the software supply chain, researchers found that attackers chained together multiple tactics, techniques, and procedures (TTPs) to evade detection and poison a popular GitHub community.
Steps of the Attack
This attack highlights the increasingly sophisticated methods that threat actors use to compromise the software supply chain, including:
- Typosquatting: Python package mirror “files[.]pypihosted[.]org” that uses a misspelling of popular “files[.]pythonhosted[.]org”
- Malicious “Colorama” copy: Inserting malicious code into a copy of a popular package then hosting it on the typosquat domain
- Bypassing authentication: Using stolen session cookies to gain access to GitHub accounts
- Leveraging reputable GitHub accounts: Taking over accounts, like editor-syntax who maintains Top.gg GitHub, to insert instructions into the repository to download the Colorama copy
- Evading detection: Committing multiple files that include malicious link and legitimate files to blend in with legitimate dependencies so users would be less likely to identify it during manual review
Understanding the Malicious Package
To spread the malware and remain hidden, the attackers manipulated the packaged installation process and the trusted Python package ecosystem.
To dig a little deeper into the attack, you should understand how the attacker embedded the malicious package into the Python fetch and execute process:
- User downloads the component containing the typosquatted, fake “colorama” from files[.]pypihosted[.]org that include malicious code located in either colorama/tests/__init__.py or colorama/init.py
- Malicious payload is hidden by using whitespace so anyone engaging in manual inspection needs to scroll horizontally for a long time.
- Malicious component fetches and executes:
- Code from “hxxps[:]//pypihosted[.]org/version” which installs additional components and decrypts hard-coded data using “fernet” library
- Code saved in a temporary file that a legitimate Python interpreter executed
- Malicious component fetches more hidden malicious code from “hxxp[:]//162[.]248[.]100[.]217/inj” then executes it.
- Code selects a folder and file name on the compromise host then retrieves the final component from “hxxp[:]//162[.]248[.]100.217[:]80/grb.”
- Malicious code establishes persistence by modifying the Windows registry, ensuring that system reboots execute the code.
Techniques used to obfuscate the malicious code include:
- Character strings containing Chinese and Japanese
- zlib compression
- Misleading variable names
The attackers use five different usernames, each associated with different malicious packages:
- Username: pypi/xotifol394
- jzyrljroxlca
- wkqubsxekbxn
- eoerbisjxqyv
- lyfamdorksgb
- hnuhfyzumkmo
- hbcxuypphrnk
- dcrywkqddo
- Username: pypi/poyon95014
- mjpoytwngddh
- Username: pypi/tiles77583
- eeajhjmclakf
- Username: pypi/felpes
- yocolor
- coloriv
- colors-it
- pylo-color
- Username: felipefelpes
- type-color
Indicators of compromise (IoCs) currently include the following:
- hxxps[:]//files[.]pythanhosted.org/packages/d8/53/6f443c9a4a8358a93a6792e2acffb9d9d5cb0a5cfd8802644b7b1c9a02e4/colorama-0.4.5.tar.gz
- hxxps[:]//files[.]pypihosted.org/packages/d8/53/6f443c9a4a8358a93a6792e2acffb9d9d5cb0a5cfd8802644b7b1c9a02e4/colorama-0.4.6.tar.gz
- hxxps://files[.]pypihosted[.]org/packages/d8/53/6f443c9a4a8358a93a6792e2acffb9d9d5cb0a5cfd8802644b7b1c9a02e4/colorama-0.4.3.tar.gz
- 162[.]248.101.215
- pypihosted.org/version
- 162[.]248.100.217
- 162.248.100.117
- 0C1873196DBD88280F4D5CF409B7B53674B3ED85F8A1A28ECE9CAF2F98A71207
- 35AC61C83B85F6DDCF8EC8747F44400399CE3A9986D355834B68630270E669FB
- C53B93BE72E700F7E0C8D5333ACD68F9DC5505FB5B71773CA9A8668B98A17BA8
3 Takeaway Lessons
This new attack highlights some significant changes in the way threat actors seek to poison the software supply chain.
Vetting Contributors May Not Work
Developers are always told to – and should always be engaging in – contributor reviews prior to using an open-source component. This attack undercuts that mitigation strategy by leveraging stolen credentials to exploit known, respected contributors.
Even after reviewing repositories, you should ensure that you:
- Identify all components and their dependencies
- Document components and dependencies using a Software Bill of Materials (SBOM)
Manual Code Reviews Miss Things
By using whitespace and horizontal scrolling to hide their malicious code, the attackers are hoping that developers manually review code rather than using an automated solution. To mitigate risks, you should:
- Scan your code regularly to identify vulnerabilities
- Compare unknown libraries against open-source and previously analyzed libraries to identify security issues
- Integrate security into your existing CI/CD pipelines, ticketing systems, and development tools
Continuously Monitor to Detect Changes
This attack highlights the way threat actors seek to compromise respected contributors and use known packages as a deployment tool. As part of building security into your development processes and application, you should ensure that you can:
- Identify changes related to dependencies
- Compare scans over time to identify changes that might otherwise go unnoticed
- Review apps and dependencies when new IoCs are identified
Qwiet AI: Securing Applications at the Source Code
With Qwiet AI, you can integrate security testing into your current CI/CD pipelines, ticketing systems, and development tools. By building security directly into your current processes, our platform enables you to incorporate container security into your secure software development life cycle (SSDLC) processes while still ensuring that you get the speed you need to deliver software on time.
The Qwiet AI platform gives you visibility into the context around vulnerabilities so that you can effectively prioritize remediation actions based on whether attackers can exploit a weakness in your application and account for whether attackers are currently exploiting that vulnerability in the wild.
Take our preZero platform for a free spin or contact us today to see how Qwiet AI can help you
