Dependency attacks 2026: The confidence gap
Perceived versus actual preparedness against dependency attacks
We surveyed 400 platform and security engineers on how ready they are for dependency attacks. Most feel prepared, but their controls tell a different story. This report exposes the confidence gap and how to close it.
About this guide
Dependency attacks land on a near-daily basis, and the latest wave executes malicious code at install time – before a package is reviewed, and often before an advisory exists. Attackers target trusted publishing identities and public registries because so much of modern software is assembled from open source, and AI agents pulling dependencies with little human review expand the attack surface.
We found that most teams believed they could catch a compromised package before an advisory exists, or identify one already in their environment within a day of a new advisory being issued. But adoption of the controls that do that in practice, like cooldown periods and scanning at ingestion, lagged behind that confidence.
The data revealed an architecture problem where controls run too late, after a package is already in your environment. The fix is to implement proactive controls at ingestion. This report explores what those controls look like and how to use them.
What you'll learn:
- Why conventional SCA and SAST scanners miss threats like Shai-Hulud, axios, and LiteLLM, because of where they run and not because of malicious source code.
- Where confidence and capability diverge: 73% trust their tooling to stop an install-time attack before an advisory exists, yet only 38% scan before ingestion and only 24% enforce cooldown policies automatically.
- The proactive controls that close the confidence gap: enforcing policy at ingestion, cooldown periods, and golden paths that make the secure development path the default.
