Poisoned Pipeline / Techniques / T0107
Discover internal artifacts names
This technique involves searching for internal artifact names used in an organization's container images and SCM repositories.
Internal artifact names may include names of applications, libraries, and components used in the organization's container images or code repositories.
An attacker can use this technique to identify the types of technologies and software components used by an organization, which can help them to identify potential vulnerabilities and attack vectors.
By discovering internal artifact names, an attacker can also gain insight into the organization's software development practices, and potentially identify development teams or individuals that are responsible for creating and maintaining specific applications or components.
Where this idea came from
Birsan dependency confusion; TXTBOOK/T-Bank private namespace reconstruction; conference: Black Hat USA 2026 “Turning Enterprise Update Servers Into Backdoor Factories (0_o)” treats internal artifact/update infrastructure as the target.
- https://medium.com/@alex.birsan/dependency-confusion-4a5d60fec610
- https://www.blackhat.com/us-26/briefings/schedule/index.html
Mitigations
- M1001 Avoid Predictable Naming Conventions
- M1503 Implement SCA analysis
- M1732 Implement code scanning for security risks
Detections
- D1260 Implement regular security audit and review
- D1500 Configure monitoring of used artifacts and open-source libraries
- D1510 Implement Intrusion Detection System and anti-malware
Observed in incidents (3)
- PyTorch-nightly torchtriton dependency confusion
- Dependency confusion (Alex Birsan)
- Unclaimed packages in official llms.txt files