Answers are grounded in this paper and cite their sources.
Modeling study demonstrates automated attack surface quantification in microservice delivery pipelines, highlighting pre-deployment exposure reduction.
Microservice architectures decompose an application into independently deployable services connected through network interfaces, service identities, data stores, and third-party components.This distribution changes the security boundary: each externally reachable API route, permitted service-to-service path, workload privilege, secret, and dependency can become an entry point or a propagation path.Systematic reviews of microservice security associate the architectural model with a larger and more dynamic attack surface, while run-time studies emphasize that frequent deployment and limited isolation complicate static security control [1,3].CI/CD intensifies this problem because source repositories, build runners, artifact registries, deployment identities, infrastructure-as-code, and external dependencies participate in the same delivery chain.NIST SP 800-204D and the OWASP CI/CD risk model therefore treat artifact provenance, access control, pipeline execution, credential hygiene, configuration, and visibility as security properties that have to be verified throughout the pipeline rather than only after deployment [4,5].The purpose of the paper is to develop an adaptive mechanism that minimizes the attack surface of a microservice release during CI/CD by measuring deployment exposure at each security gate, comparing the candidate state with an accepted baseline, and automatically changing configuration before promotion.The mechanism is not limited to vulnerability counting.It combines five exposure dimensions: external API reachability, excessive east-west connectivity, workload privilege, secret scope, and unresolved high-risk software dependencies.This design follows the principle that attack-surface measurement should describe the resources and channels available to an adversary, while the release decision should react to changes in those resources [6].For a candidate deployment at time t, the system state is represented by a service graph Gₜ = (Vₜ, Eₜ) and a normalized attack-surface vector xₜ = (Eₜ, Rₜ, Pₜ, Sₜ, Dₜ).Eₜ denotes the share of externally reachable routes that are not required by the declared public interface; Rₜ denotes excess service-to-service reachability relative to the declared communication graph; Pₜ is the share of workloads with privileges above the predefined execution profile; Sₜ is the share of secrets whose scope or lifetime exceeds policy; Dₜ is the normalized amount of unresolved high or critical dependency exposure.Each component is bounded to [0, 1].The aggregate Attack Surface Index (ASI) is calculated as follows (1): ASI t = w T x t (1)
No takes yet. Share an insight, caveat, or question.
Artem Rozhko (2026) studied this question.