AT-03 Role-based Trainingat-3
Whether an AWS call fetches evidence for this control, and what to run. Its Rev5 baselines, the indicators that reach it and the FedRAMP guidance attached to it open below.
No AWS API proves itA Key Security Indicator reaches AT-03 and no AWS API decides it. This project has assessed it and closed it: the evidence is a document a person writes and an assessor reads, not telemetry. The reason is below.
Collect evidence (0)
Authored AWS and pipeline recipes whose output is evidence for AT-03. This mapping is this project’s opinion (AWS overlay v3.0.0, pipeline overlay v0.8.0), versioned separately from the dataset — the upstream FedRAMP rules name none of these tools.
Role-based training records map people to courses in an LMS; the AWS role a principal assumes says nothing about what they were taught.
This project’s assessment (automation register v0.9.0, reviewed 2026-08-04), versioned separately from the dataset and from the evidence overlays. It is an opinion about the AWS surface on that date, and the surface moves.
No recipe reaches it through its indicators either. The KSI that reaches it is KSI-CED-RAT — which is what asks for it. The evidence it asks for is written, not fetched. Browse /collect for the whole authored corpus.
What NIST requires of AT-03
The control statement from NIST 800-53 Rev5, verbatim. Square brackets are organization-defined parameters — yours to set, not FedRAMP's to dictate.
a. Provide role-based security and privacy training to personnel with the following roles and responsibilities: [organization-defined roles and responsibilities]: 1. Before authorizing access to the system, information, or performing assigned duties, and [frequency] thereafter; and 2. When required by system changes; b. Update role-based training content [frequency] and following [events]; and c. Incorporate lessons learned from internal or external security incidents or breaches into role-based training.
NIST SP 800-53 Rev5 catalog 5.2.0, from usnistgov/oscal-content at 78650f0. The FedRAMP dataset carries no control text; this is borrowed and pinned.
Rev5 baseline membership
Membership is the whole relationship — a baseline is a set of control ids. Class A carries no baseline at all. The Low/Moderate/High labels are an interpretation from control counts, not a dataset fact.
KSI indicators reaching this control (1)
Under 20x, these indicators are how this control is demonstrated — automated KSI evidence stands in for narrative control evidence.
Pages exist for the 209 controls reached by at least one KSI. Baseline-only controls are the orphans on /coverage. Control titles and full text live in the NIST catalog, not this dataset.