Model and Agent Release Registry

Data storeGovernance & ComplianceSafety, Security & Governancearc:ModelRegistry

A versioned registry of deployable agent artifact sets (configuration, prompt templates, tool configurations, evaluation metrics) tagged with commit SHA and lifecycle stage for lineage and rollback.

Responsibility. Records every deployable agent version with its lineage and stage.

Also known as: Model registry, Agent version registry, Agent artifact registry, MLflow Model Registry, Agent Artifact Registry

Direct neighbourhood (hover for relationship types)

Relationships

is read by dependency

Continuous Integration Runner Ch4.4
Model Card Generator Ch9.8
Rollout Manager abstract Ch4.2
Version Rollback Controller abstract Ref8.08

is written by dependency

Design guidance

SHOULD log configuration, prompt templates, tool configurations and evaluation metrics per version with the git commit SHA as identifier.
SHOULD keep registry stage consistent with what is actually deployed by promoting versions after successful rollout.
SHOULD roll back by redeploying the previous registered version rather than reverting and rebuilding.
MUST version the full artifact constellation (model version, prompts, tool configurations, retrieval settings, evaluation dataset version), not only code or weights.
SHOULD link each version to its originating run, commit and evaluation dataset version so quality changes can be attributed to agent changes rather than test-set changes.
SHOULD be the source deployments resolve the production version from, rather than a mutable 'latest' tag.
SHOULD record post-deployment production metrics against the version alongside pre-deployment results.

Quantitative guidance

As stated by the sources; verify before use.

Registry rollback ~30 s vs ~15 minutes via git revert and rebuild (Ch4.2).
Lineage lets teams revert a faulty prompt version within minutes rather than hours of reconstruction (Ch4.4).

Classification

Patterns: Lineage trackingStage promotion (staging -> production)Registry-based rollbackLifecycle stages (None, Staging, Production, Archived)Registry-as-source-of-truth for deployment
Technologies: MLflow Model RegistryAmazon S3Azure Blob Storage
Quality attributes: Transparency and accountability (NIST AI RMF: accountable and transparent)Reliability (ISO/IEC 25010 | NIST AI RMF: valid and reliable)
Risks mitigated: Slow rollback by rebuilding from gitInability to trace a production incident to a commitInability to identify what changed between working and failing deploymentsUntraceable production decisions in regulated settings
Frameworks & regulations: ISO/IEC 42001 §8 Operation (change control)

Sources

Ch4.2: T. Nguyen, "Deployment and Scaling," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 4.2. ISBN: 9798244538229.
Ch4.4: T. Nguyen, "Performance Profiling and Optimization," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 4.4. ISBN: 9798244538229.
Ch9.8: T. Nguyen, "Standards and Frameworks for AI Governance," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 9.8. ISBN: 9798244538229.