CSPM
Tooling that continuously reads cloud control-plane configuration and flags resources that break policy, such as public buckets or open security groups.
CSPM polls cloud provider APIs (or consumes config change events) to build an inventory of every account, resource and setting, then evaluates that inventory against rules: CIS Benchmarks, internal policy, compliance frameworks. Output is a list of findings tied to a resource and an owner.
It exists because cloud misconfiguration, not novel exploits, causes most cloud breaches, and humans cannot review thousands of accounts by hand. Good platforms go beyond single-resource checks and build an attack path graph: a public load balancer to an instance with an over-privileged role that can read a sensitive bucket is one finding, not three.
Where it goes wrong: alert volume with no ownership, API rate limits that make scans stale, and treating CSPM as prevention. It is detective; prevention belongs in guardrails (SCPs, org policies, admission and IaC checks in CI). In interviews, expect to design one: event-driven ingestion plus periodic full sweeps, multi-tenant isolation and drift detection.