Supply chain security for financial services

Could a malicious package already be running in your systems?

Open-source packages execute install scripts the moment they land - reaching systems that handle financial data and customer accounts before your scanner ever runs. Cloudsmith scans and blocks at ingestion, before packages arrive.

Software isn't written anymore. It's assembled.

Modern applications are built from hundreds of open-source dependencies, most several layers removed from any code your team wrote directly. AI-assisted development has accelerated that pattern further. AI-generated code draws on the same open-source dependencies as code written by hand, often introducing packages that no engineer reviewed.

Open source and AI tooling are the reason why modern engineering teams move as quickly as they do. The instinct in a regulated environment can be to respond to that pace with restriction – lock down package sources, ban AI-assisted tooling, slow everything to a pace security can manually review. That answer trades velocity for safety, when the real gap is at the point of ingestion, not at the point of adoption.

Existing security postures have a timing problem

The standard model for dependency security pulls a package from a registry, scans it, then promotes it to developers and pipelines. That sequence treats a scan result as a safety guarantee, but the scan only runs after the package has already had its chance to do damage.

Packages across npm, PyPI, and other ecosystems can execute code the moment they're installed or built, not just when your application runs – often inside your build infrastructure, with access to the API keys, cloud credentials, and environment variables your pipelines depend on.

Closing one execution path doesn't close the category. Ecosystems are tightening default behaviour around install-time scripts, but attackers are already shifting to build steps and native compilation triggers that fall outside what's being restricted – the mechanism changes faster than the restrictions can keep up.

Cloudsmith works differently. Policy evaluation runs at ingestion, before the pull completes, so a package that fails policy never reaches your environment in the first place.
  • AI-era protection: AI agents pulling dependencies at speed amplify the risk. Cloudsmith inspects every package they request, applying the same governance rules as for human developers.
  • Automated quarantine: Packages that breach your CVE thresholds, license rules, or soak-period policies are held automatically and never reach your developers.
  • Enforcement across every package: Every dependency, direct or transitive, is evaluated against rules that you define before it reaches your engineers or your customers, giving your teams a governed path to open source without restricting access or delaying work.

Compliance that doesn't require manual overhead

Financial services teams carry specific compliance obligations – data residency requirements, artifact retention policies, audit trails for regulatory review. Managing these inside self-hosted infrastructure is a meaningful and avoidable cost.

Cloudsmith handles artifact retention with configurable rules. Set the policy once and packages are retained or expired automatically – retention periods beyond seven years are supported. Data residency controls let you specify where artifacts are stored. Every pull, push, and policy decision is logged and exportable.

DORA Article 28 requires documented management of third-party ICT risk. An audit trail of what entered your environment, when, and what policy it was evaluated against is evidence of that control. Cloudsmith's logs provide that evidence.

For organizations operating across multiple teams or acquired entities, Cloudsmith provides a single artifact store across all package formats – Maven, npm, PyPI, Docker, Helm, and more.

What happens after the disclosure

When a package is compromised, your team needs answers fast. Security needs to know whether you were exposed. Compliance needs evidence for regulators. Customers need a response before they ask for one. These are the questions that decide how quickly you can answer.

Do we use this package?

Without comprehensive SBOMs, it can be impossible to know whether a compromised package has slipped in as a transitive dependency. Cloudsmith automatically generates SBOMs and keeps them current, ensuring they are always available to answer questions.

Did we pull the impacted version?

Identifying which version of a package you are using can become a long, manual process without a full record of every request made by your developers.

Where was it deployed?

If a breach occurs, your customers will need immediate answers about whether they were impacted. You may also face reporting obligations under laws such as the EU's Cyber Resilience Act (CRA). A complete log of every package you pulled is the quickest way to get these answers and reassure your customers.
The threat landscape

Financial infrastructure is a target. Your dependencies are the vector.

Attackers have compromised npm and PyPI packages used directly by financial and crypto-adjacent engineering teams in the past six months, stealing wallet credentials, cloud tokens, and SSH keys from systems that process real transactions before any scanner had a chance to identify a malicious package.
The Miasma worm's path of destruction

The Miasma worm's path of destruction - June 8, 2026

What started as an exploit in Red Hat’s npm packages has escalated to a sprawling supply chain disaster, spreading to 73 Microsoft GitHub repos across popular environments like Microsoft Azure and Durable Task. The fallout of this worm further emphasizes how the software security landscape has drastically shifted, creating new risks for vendors and their customers.
Axios 1.4.1 and 0.30.4 NPM packages compromised

Axios 1.4.1 and 0.30.4 NPM packages compromised - Mar 31, 2026

In the early hours of 31 March 2026, axios, an extremely popular HTTP client for JavaScript, suffered a malicious payload injection attack. While the malicious code was removed within a matter of hours, the package’s popularity means that millions of build systems may have been infected, halting development at vendors and placing their customers at risk.
How Cloudsmith can protect against the LiteLLM attack

How Cloudsmith can protect against the LiteLLM attack - May 12, 2026

A backdoored LiteLLM release exposed SSH keys, cloud credentials, and API keys across millions of developer environments in March 2026. Cloudsmith's artifact management controls can block this class of attack before it reaches your team or your customers.

Take the Security Maturity Assessment

See where your current security posture stands against the realities of software development in 2026.

See Cloudsmith in action

Watch how Cloudsmith protects enterprise pipelines from supply chain threats, at any scale.

Talk to our team

See how Cloudsmith secures and scales artifact management for software vendors, helping them to protect their customers. Book a demo with our team.

Frequently asked questions

Common questions from software vendors evaluating Cloudsmith. Got a question not covered here? Reach out to our team.

  1. Yes, and we make the transition as smooth as possible. Our Ultra and Enterprise plans include full onboarding support and a dedicated customer success manager who works with your team through migration planning, data transfer, and pipeline reconfiguration. We have helped dozens of enterprises migrate from JFrog Artifactory, Sonatype Nexus, and homegrown solutions without disrupting active build pipelines.

  2. Cloudsmith evaluates every package at the point of ingestion, before it enters your repositories. This is the critical difference from tools that scan after the fact. When a package is flagged, Cloudsmith can quarantine it automatically, block promotion to production environments, and alert your security team, all governed by policies you define in OPA Rego. New CVE disclosures trigger continuous re-evaluation of packages already in your repositories, so a clean package today does not become a silent liability tomorrow.

  3. Yes. Cloudsmith supports 30+ package formats natively, including Docker, npm, Maven, PyPI, NuGet, Helm, Debian, RPM, Cargo, and more. All formats are managed through a single control plane with consistent security policies, access controls, and audit logging. You do not need separate tooling or governance processes for different ecosystems.

  4. Cloudsmith is cloud-native with elastic scaling, so there is no capacity ceiling. Storage and bandwidth scale automatically as your artifact volumes grow. Our global edge network spans 600+ points of presence, ensuring fast artifact delivery to engineering teams and CI/CD runners anywhere in the world. We back this with a 99.99% uptime SLA with multi-region redundancy, so your build pipelines are never blocked by infrastructure issues.

  5. Cloudsmith's Enterprise Policy Manager uses OPA Rego to define security and compliance rules as code. Your policies are version-controlled, auditable, and applied consistently across every repository, team, and package format. You can enforce CVE thresholds, license restrictions, package age (soak periods), and custom rules based on package metadata, all automated, with no manual gatekeeping required.

  6. Cloudsmith maintains a complete, immutable audit trail of every artifact ingested, every package pulled, every policy decision made, and every user action taken across your organization. When a security incident occurs, your team can trace the blast radius quickly, identifying exactly which packages, pipelines, and environments were affected. This log data can be exported to your SIEM or observability platform for centralized monitoring.

  7. Cloudsmith integrates with your Identity Provider via SAML/SSO and SCIM. User provisioning and deprovisioning happen automatically when you add or remove users in your IdP, so there is no manual offboarding. OIDC support replaces static API keys with short-lived tokens in your CI/CD pipelines, eliminating standing credential risk. Role-based access control lets you define precise permissions across teams, repositories, and environments.

  8. Cloudsmith is designed for the AI-enabled engineering era. When AI agents write code and install dependencies at high speed and volume, your artifact management platform needs to keep pace while maintaining governance. Cloudsmith inspects every dependency AI agents request against your policies, applying the same controls as for human developers. AI-generated software does not escape your security guardrails simply because it was produced by an agent.