| name | review-doc-alignment |
| metadata | {"author":"rhoai3-coding-demo","version":"1.0.0","platform-family":"rhoai","platform-baseline":"repo","ocp-baseline":"repo","skill-group":"GitOps & Manifests"} |
| description | Verify that GitOps manifests align with official Red Hat documentation for the platform baseline. Use when creating or modifying CRs, operator configurations, InferenceServices, ServingRuntimes, or any RHOAI-managed resource. Also use for periodic alignment audits across stages. Do NOT use for label/YAML compliance (use review-manifest-compliance), live cluster troubleshooting (use inspect-cluster), or workflow-level reviews (use review-gitops-change).
|
Review Documentation Alignment
Verify that manifests match the official Red Hat documentation — not just that they are valid YAML.
When to invoke
- Creating or modifying CRs, operator configurations, or RHOAI-managed
resources
- Periodic alignment audit across stages
- After platform baseline version changes
Preparation
- Read
docs/PLATFORM_BASELINE.md to confirm the active product versions.
- Read
.agents/rules/rhoai.md and .agents/rules/ocp.md for domain
guardrails.
- Consult the relevant
rhoai-* or ocp-* skills for the resource type.
What to check for each manifest
1. API version correctness
- Is the
apiVersion the one documented for this resource in the active
platform baseline version?
2. CR field validity
- Are all spec fields documented for this CR version?
- Are there fields that look invented or copied from a different version?
- Consult the matching
rhoai-* or ocp-* skill for the resource kind and
verify each top-level spec field.
3. Operator configuration
- Do Subscription channels match what the platform baseline version expects?
- Are operator names and catalog sources correct?
4. Annotation correctness
- Dashboard annotations (
opendatahub.io/template-name, etc.) — do the values
match actual platform templates?
- ArgoCD annotations — are sync-wave values reasonable for the stage order?
5. Image references
- Do container images reference Red Hat registry (
registry.redhat.io) or
approved sources?
- Are image tags pinned (not
:latest for production-grade components)?
6. Referenced documentation
- Does the manifest's README reference the correct doc section for the active
platform baseline?
- Are the doc links still valid (not pointing to older versions)?
How to review a stage
- Read all YAML files in
gitops/stages/NNN-name/base/
- For each CR, consult the matching
rhoai-* or ocp-* skill and compare
fields
- For each operator Subscription, verify the channel and source against docs
- Check the stage README References section for correct doc links
- Report findings
Key RHOAI resource types
| Resource | Skill to consult |
|---|
| DataScienceCluster | rhoai-dsci-dsc-configuration |
| DSCInitialization | rhoai-dsci-dsc-configuration |
| HardwareProfile | rhoai-hardware-profiles |
| InferenceService | rhoai-model-deployment |
| ServingRuntime | rhoai-model-serving-platform |
| ModelRegistry | rhoai-model-registry |
| DataSciencePipelinesApplication | rhoai-ai-pipelines |
Output format
For each stage reviewed:
Stage: NNN-name
Files reviewed: N
Doc-Aligned:
- subscription.yaml: channel matches platform baseline docs
- datasciencecluster.yaml: spec fields match installation guide
Misaligned:
- [API] guardrails.yaml: apiVersion should be X per docs, found Y
- [FIELD] config.yaml: field 'foo' not documented
- [DOC-REF] README.md: references a different version than platform baseline
Summary: X aligned, Y misaligned
Related skills
review-manifest-compliance — label, selector, YAML standards compliance
review-gitops-change — workflow-level review with security and MaaS impact
inspect-cluster — gather live cluster state