TL;DR: Malicious package releases, release-asset swaps, and runtime backdoors are now stealing AWS, Azure, GCP, Kubernetes, Vault, SSH, and AI-tool credentials across Python, npm, Go, NuGet, and CMS ecosystems, according to Corgea. The operational boundary has shifted from dependency hygiene to full secret containment, because poisoned packages can trigger credential harvesting, exfiltration, and lateral movement before detection.
NHIMG editorial — based on content published by Corgea: weekly briefing on supply-chain, kernel, and application-security findings
By the numbers:
- 79% of organisations have experienced secrets leaks, with 77% of these incidents resulting in tangible damage.
- 91.6% of secrets remain valid five days after the targeted organisation is notified, showing a critical gap in remediation procedures.
Questions worth separating out
Q: What breaks when a trusted package can run code on developer or CI systems?
A: The main failure is that software provenance does not stop runtime abuse.
Q: Why do cloud and AI-tool credentials increase the blast radius of supply-chain attacks?
A: They often bridge multiple systems, so one stolen secret can unlock storage, compute, source control, vaults, and automation.
Q: What do teams get wrong about dependency provenance and package trust?
A: Teams often assume a signed or expected package name is enough, but package metadata, lifecycle hooks, and publishing path all matter.
Practitioner guidance
- Harden package execution paths Restrict where untrusted dependencies can execute, including install hooks, import-time code, and CI runner activation paths.
- Rotate exposed secrets from clean systems Assume any host that imported a malicious package may have leaked cloud, vault, SSH, or AI-tool credentials.
- Scope developer and CI credentials tightly Use short-lived credentials with minimal permissions for build systems, package publishing, and automation accounts.
What's in the full report
Corgea's full weekly briefing covers the operational detail this post intentionally leaves for the source:
- Package-by-package findings, including affected versions and the exact malicious behaviours observed during execution.
- Cross-source reporting notes from Aikido, Endor Labs, SafeDep, and other investigators behind the week’s disclosures.
- Specific remediation advice for teams that need to triage Linux hosts, CI runners, and developer workstations after exposure.
- Additional coverage of the Strapi, Nx Console, and Node IPC issues that broaden the week’s supply-chain risk picture.
👉 Read Corgea's weekly briefing on supply-chain, kernel, and application-security findings →
Supply-chain worm credential theft: what IAM teams need to know?
Explore further
Trusted package provenance is not a sufficient security boundary. Provenance signals help with software authenticity, but they do not stop a signed or legitimately published artifact from behaving maliciously after install. In supply-chain incidents like this, the relevant trust boundary is runtime behaviour, not publication lineage. Teams should therefore evaluate execution permissions, egress paths, and secret reachability as first-class controls.
A few things that frame the scale:
- 91.6% of secrets remain valid five days after the targeted organisation is notified, showing a critical gap in remediation procedures, according to Ultimate Guide to NHIs.
- 97% of NHIs carry excessive privileges, increasing unauthorised access and broadening the attack surface, according to Ultimate Guide to NHIs.
A question worth separating out:
Q: Who is accountable when a malicious dependency steals secrets from CI runners?
A: Accountability usually spans engineering, platform, security, and the team that approved the publishing or build workflow. The control gap sits across software supply-chain governance, secrets management, and identity lifecycle discipline. Frameworks such as NIST CSF, NIST SP 800-53, and OWASP NHI all touch different parts of that shared responsibility.
👉 Read our full editorial: Supply-chain worms are now stealing cloud and AI credentials