Most security plans, control baselines and assessment results still live in Word and Excel documents that no tool can read, so every audit and authorisation starts with someone retyping evidence. OSCAL, the Open Security Controls Assessment Language led by NIST, replaces those documents with structured XML, JSON or YAML that tools can validate, exchange and update continuously. This guide explains the seven OSCAL models and how they reference each other, what the FedRAMP machine-readable package proposal means for cloud providers, how to describe non-human identities in an OSCAL system security plan, and how to roll OSCAL out without rebuilding your compliance programme. It is written for compliance engineers, GRC teams, assessors and cloud providers moving to compliance-as-code. For the wider automation picture see the Compliance Automation Guide; this page goes deep on the data standard.
Key takeaways
- Seven models in three layers. Catalog and Profile (control layer), Component Definition and System Security Plan (implementation layer), and Assessment Plan, Assessment Results and Plan of Action and Milestones (assessment layer), all released in XML, JSON and YAML.
- Each layer points to the one below. A profile imports controls that must originate in a catalog, an SSP is written against a profile, and assessments are expressed against a specific system implementation.
- NIST already publishes the catalogs. SP 800-53 Rev. 5.2.0 (with SP 800-53A procedures), SP 800-171 Rev. 3 and CSF 2.0 are available as OSCAL catalogs, so nobody needs to transcribe controls.
- FedRAMP is the forcing function. FedRAMP's RFC-0024 proposed that new Rev5 packages be machine-readable from 30 September 2026, with OSCAL on the approved format list; check FedRAMP's current rules for the final position.
- Machine identities become data, not prose. OSCAL SSPs describe users, roles, privileges and components as structured objects, which lets you list service accounts and API integrations once and reuse them across controls and assessments.
What it is and who it applies to
NIST describes OSCAL as an initiative developed with industry to modernise and automate security and compliance processes, moving from document-based plans to data standards. Typical uses NIST lists include turning regulatory policy into machine-readable policy-as-code, sharing control baselines, keeping current records of how controls are implemented, and automating monitoring of control effectiveness.
The latest release on the NIST OSCAL Reference site is v1.2.3 (checked 27 October 2026). The OSCAL version is recorded inside every document in the metadata.oscal-version field, which is how tools decide which schema to validate against.
- Cloud service providers selling to US federal agencies: FedRAMP packages (SSP, assessment plan, assessment report and POA&M) are the main driver. See the FedRAMP Guide.
- Federal agencies and assessors: agencies ingesting authorisation data and assessors producing results benefit from validated, comparable data.
- Compliance engineering teams anywhere: any organisation mapping several frameworks can use catalogs, profiles and component definitions to avoid duplicating control narratives.
- Product vendors: component definitions let a vendor publish how its product supports specific controls once, for every customer to import.
Structure and key requirements
The seven models
| Layer | Model | What it holds | Points to |
|---|---|---|---|
| Control | Catalog | Controls from any control framework, with parameters, statements, guidance and, for SP 800-53, assessment objectives and methods | Nothing (the base) |
| Control | Profile | A selection and tailoring of controls, such as a baseline: which controls, parameter values, additions and modifications | One or more catalogs or other profiles |
| Implementation | Component Definition | How a component (policy, hardware, software, service) implements controls; used to pre-populate SSPs | Catalogs or profiles |
| Implementation | System Security Plan (SSP) | A system's description, boundary, users, components and how each control is implemented | A profile (the baseline) |
| Assessment | Assessment Plan | How, when and in what scope an assessment will be performed | The SSP |
| Assessment | Assessment Results | What was performed, evidence collected and findings | The assessment plan |
| Assessment | Plan of Action and Milestones (POA&M) | Findings the system owner must address, with remediation tracking | The SSP and assessment findings |
The OSCAL Reference site also lists a Control Mapping model in v1.2.3 for expressing relationships between controls in different frameworks, and a prototype shared responsibility model. Treat these as newer and less widely supported by tools than the seven core models.
Anatomy of a catalog
The NIST SP 800-53 OSCAL catalog in NIST's oscal-content repository is a good reference. It is titled "Electronic (OSCAL) Version of NIST SP 800-53 Rev 5.2.0 Controls and SP 800-53A Rev 5.2.0 Assessment Procedures", declares oscal-version 1.2.3 and contains 20 control families as groups holding 1,196 controls and enhancements (withdrawn items included). Each control object has id (for example ia-5), title, params, props, links, nested controls for enhancements, and parts named statement, guidance, assessment-objective and assessment-method. A back-matter section holds referenced resources.
Anatomy of an SSP
An OSCAL SSP's root object has a uuid, metadata (title, version, oscal-version, roles, parties, responsible-parties and more), import-profile (an href to the baseline), system-characteristics (system IDs and name, description, authorisation date, sensitivity level, information types and impact levels), system-implementation, control-implementation and back-matter. In system-implementation, users carry role IDs and authorised privileges, and components and inventory items describe what is in the boundary. In control-implementation, each implemented requirement is tied to a control ID and broken down by component, which is what lets a single component description serve many controls.
Validation rules
- Well-formed and valid: NIST states OSCAL tools must check that documents are well-formed and valid against the schema for their version.
- Referential integrity: every control a profile imports must originate in a catalog, so broken
hrefor control IDs fail resolution. - Stable identifiers: UUIDs identify objects across revisions; regenerating them on every export breaks history and cross-references.
Identity and NHI requirements
OSCAL does not add identity controls of its own; it changes how you describe them. In a Word SSP, service accounts appear as sentences scattered across AC-2, IA-5 and AC-6 narratives. In OSCAL they can be modelled once and referenced everywhere.
| Identity element | Where it lives in OSCAL | Good practice |
|---|---|---|
| Human and non-human user types | SSP system-implementation users, with role IDs and authorised privileges | Model service accounts, CI/CD identities and AI agents as distinct user entries with their privileges, not folded into "administrators" |
| Identity providers, vaults and PAM tools | Components (and component definitions supplied by the vendor) | Reference the secrets manager component from every control it supports (IA-5, SC-12, AC-6) |
| Interconnections and API integrations | Components describing external services and interconnections | Record the credential type each integration uses and who owns it |
| Control parameters | Profile set-parameters or SSP parameter settings | Set concrete values, for example authenticator lifetime for IA-5 or account review frequency for AC-2 |
| Assessment evidence | Assessment Results observations and findings | Attach machine-generated evidence such as account inventories and key rotation reports rather than screenshots |
The payoff is consistency: when a new agent integration is added, one component and one user entry change, and every affected control picks it up. It also exposes gaps, because a privileged machine user with no linked component or control implementation is visible in the data. See the Service Account Inventory Template for the fields worth capturing, and the Agent Identity and Access Guide for agent-specific privileges.
Implementation steps
- Start from published content. Use NIST's OSCAL catalogs and published baselines rather than authoring your own; record the exact catalog version (for example SP 800-53 Rev 5.2.0).
- Pick one serialisation and one OSCAL version. JSON is the usual choice for pipelines; XML is common with assessors. Pin the OSCAL version and validate everything against its schema in CI.
- Build or import the profile. Select the baseline, set organisation-defined parameters and resolve the profile to confirm every imported control exists.
- Write component definitions. Describe shared components (identity provider, secrets manager, logging platform, CI/CD) once, with the controls each supports. Ask vendors whether they publish OSCAL component definitions.
- Assemble the SSP from components. Generate the SSP from system inventory and component definitions instead of hand-writing narratives; include every non-human user type with its privileges.
- Pilot with one system. Convert one existing SSP, compare it line by line with the document version, and fix modelling decisions before scaling.
- Automate evidence into Assessment Results. Feed scanner, configuration and identity inventory outputs into observations so findings carry their evidence.
- Drive the POA&M from findings. Generate POA&M items from assessment findings and update status from ticketing data.
- Keep a human-readable view. Render documents for reviewers who still need them; FedRAMP's RFC-0024 proposal itself expected Word or Excel versions on request.
- Plan rollback. Keep the document-based process available until assessors and agencies have accepted at least one OSCAL package end to end.
Evidence and assessment
FedRAMP machine-readable packages
FedRAMP's RFC-0024, "FedRAMP Rev5 Machine-Readable Packages", proposed the following for the Rev5 process (not 20x). Its comment period closed on 11 March 2026 and FedRAMP's RFC index listed no final outcome when checked on 27 October 2026, so treat these as proposed requirements and confirm against FedRAMP's current consolidated rules.
| Proposed requirement | Proposed date |
|---|---|
| FedRAMP accepts machine-readable packages and gives them priority review, targeting 90 days | Tentatively 15 April 2026 |
| New authorisation packages submitted in an approved machine-readable format (OSCAL on the approved list while maintained) | 30 September 2026 |
| Full machine-readable package for each annual assessment completed afterwards | After 30 September 2026 |
| Packages updated by the end of the month following a significant change | 30 September 2026 |
| Word or Excel versions provided when FedRAMP or an agency requests them | 30 September 2026 |
| Grace period ends; non-compliant providers lose certification | 30 September 2027 |
Measuring the programme
- Percentage of systems with a schema-valid OSCAL SSP in version control.
- Percentage of control implementations generated from component definitions rather than free text.
- Percentage of assessment observations backed by machine-generated evidence.
- Time from a significant change to an updated, validated package.
- Number of non-human user types in the SSP versus the identity inventory (the gap should be zero).
Mapping to other frameworks
OSCAL is a format, so it maps to the processes that consume it rather than to individual controls.
| Framework or process | How OSCAL is used | Relevant controls or outcomes |
|---|---|---|
| NIST SP 800-53 Rev. 5 and SP 800-53A | Official OSCAL catalog with assessment objectives and methods | PL-2 System Security and Privacy Plans; CA-2 Control Assessments; CA-5 Plan of Action and Milestones; CA-7 Continuous Monitoring |
| NIST RMF (SP 800-37) | Profiles for baselines, SSPs for implementation, results and POA&M for authorisation | CA-6 Authorization; PM-31 Continuous Monitoring Strategy. See the NIST Risk Management Framework (SP 800-37) Guide. |
| FedRAMP Rev5 | Machine-readable authorisation packages (proposed RFC-0024) | SSP, SAP, SAR and POA&M |
| NIST CSF 2.0 | Official OSCAL catalog | ID.IM-01 (improvements from evaluations); GV.OV-01 (strategy outcomes reviewed) |
| NIST SP 800-171 Rev. 3 and CMMC | Official OSCAL catalog for CUI requirements | See the NIST SP 800-171 and CMMC 2.0 Guide |
| Multi-framework mapping | Control Mapping model and NIST informative references | See the NIST CSF 2.0 Informative References and OLIR Guide |
Choosing how far to go
| Situation | Sensible OSCAL scope | Trade-off |
|---|---|---|
| Small provider, one system, FedRAMP pursuit | SSP, assessment artefacts and POA&M in OSCAL, generated by a GRC tool or template; reuse NIST catalogs and published baselines | Fast to adopt, but limited reuse beyond the one package |
| Large provider, many systems and frameworks | Shared component definitions, internal profiles per framework, SSPs generated from inventory, evidence pipelines into Assessment Results | Higher upfront engineering, large saving on every later audit and change |
| Organisation outside federal markets | Catalogs and profiles for multi-framework mapping first; SSPs later if an assessor accepts them | Value depends on tool support among your assessors |
Common pitfalls
- Converting Word to JSON and stopping. Pasting narrative text into
remarksfields produces valid OSCAL with none of the benefits. The value comes from components, parameters and references. - Version drift. Mixing documents from different OSCAL versions, or catalogs from different SP 800-53 releases, causes validation and resolution failures. Pin both.
- Regenerated UUIDs. Tools that mint new UUIDs on every export break the link between findings, POA&M items and the SSP.
- Machine identities missing from users. SSPs that list only human roles hide service accounts and agent privileges from assessors, and the gap shows up later as an AC-2 or IA-5 finding.
- Manual evidence. Screenshots attached as back-matter resources are no better than before; feed tool output into observations.
- No human review path. Assessors and authorising officials still need to read the package. Plan rendering from day one.
Practitioner checklist
- Pin an OSCAL version (v1.2.3 was latest when checked) and validate every document in CI.
- Import NIST's published catalogs and record the catalog release used.
- Build the baseline as a profile with explicit parameter values and resolve it.
- Write component definitions for identity, secrets, logging and CI/CD platforms.
- List every non-human user type with role IDs and authorised privileges in the SSP.
- Keep UUIDs stable across revisions.
- Route scanner and identity inventory output into Assessment Results observations.
- Generate POA&M items from findings and sync status from ticketing.
- Track FedRAMP's final position on RFC-0024 if you hold or seek FedRAMP certification.
Standards and references
- NIST: Open Security Controls Assessment Language (OSCAL)
- NIST: OSCAL Layers and Models
- NIST: OSCAL Model Reference
- NIST: OSCAL System Security Plan Model JSON Outline
- NIST: OSCAL Content repository
- NIST: SP 800-53 Rev. 5
- FedRAMP: RFC-0024 FedRAMP Rev5 Machine-Readable Packages
- FedRAMP: Requests for Comment
Related NHI Mgmt Group resources: Compliance Automation Guide · FedRAMP Guide · NIST SP 800-53 Guide · NIST SP 800-53A Guide · NIST SP 800-137 Information Security Continuous Monitoring Guide