aws · aneeshamz · Oct 4, 2026
diff --git a/cloudformation/devops-agent-skill-policies.yaml b/cloudformation/devops-agent-skill-policies.yaml
@@ -32,6 +32,7 @@ Metadata:
           - EnableAgentCoreObservabilitySetup
           - EnableAgentCoreOpsReview
           - EnableSageMakerAIOpsReview
+          - EnableAwsFsxnOperationsReview
       - Label:
           default: Optional Resource Scoping
         Parameters:
@@ -147,6 +148,16 @@ Parameters:
     Default: 'true'
     AllowedValues: ['true', 'false']
 
+  EnableAwsFsxnOperationsReview:
+    Type: String
+    Description: >
+      Amazon FSx for NetApp ONTAP Operational Review skill (read-only: adds
+      fsx:Describe*/ListTagsForResource, cloudwatch metric/alarm reads,
+      backup:ListProtectedResources/ListRecoveryPointsByResource, and
+      ec2:DescribeSecurityGroups/Subnets/RouteTables).
+    AllowedValues: ['true', 'false']
+    Default: 'true'
+
 Conditions:
   CreateNewRole: !Equals [!Ref ExistingRoleName, '']
   SkillAwsHealthEvents: !Equals [!Ref EnableAwsHealthEvents, 'true']
@@ -161,6 +172,7 @@ Conditions:
   SkillAgentCoreObservabilitySetup: !Equals [!Ref EnableAgentCoreObservabilitySetup, 'true']
   SkillAgentCoreOpsReview: !Equals [!Ref EnableAgentCoreOpsReview, 'true']
   SkillSageMakerAIOpsReview: !Equals [!Ref EnableSageMakerAIOpsReview, 'true']
+  SkillAwsFsxnOperationsReview: !Equals [!Ref EnableAwsFsxnOperationsReview, 'true']
   HasRegionRestriction: !Not [!Equals [!Join ['', !Ref AllowedRegions], '']]
 
 Resources:
@@ -527,6 +539,44 @@ Resources:
               - savingsplans:DescribeSavingsPlans
             Resource: '*'
 
+  # aws-fsxn-operations-review: strictly read-only FSx for NetApp ONTAP review.
+  # Grants the skill's full read set (fsx control-plane Describe*/ListTagsForResource,
+  # CloudWatch metric/alarm reads, backup protected-resource/recovery-point lists, and
+  # ec2 network describes for the security/networking checks). No Start*, Put*, Create*,
+  # Update*, or Delete* is granted. sts:GetCallerIdentity is intentionally omitted — it
+  # requires no IAM permission. These actions were NOT delta-trimmed against
+  # AIDevOpsAgentAccessPolicy with iam:SimulatePrincipalPolicy; a maintainer may remove
+  # any already covered by the managed policy (e.g. backup:List*, cloudwatch reads).
+  PolicyAwsFsxnOperationsReview:
+    Type: AWS::IAM::Policy
+    Condition: SkillAwsFsxnOperationsReview
+    Properties:
+      PolicyName: DevOpsAgentSkill-AwsFsxnOperationsReview
+      Roles:
+        - !If [CreateNewRole, !Ref DevOpsAgentRole, !Ref ExistingRoleName]
+      PolicyDocument:
+        Version: '2012-10-17'
+        Statement:
+          - Sid: FsxnOperationsReviewReadOnly
+            Effect: Allow
+            Action:
+              - fsx:DescribeFileSystems
+              - fsx:DescribeVolumes
+              - fsx:DescribeStorageVirtualMachines
+              - fsx:DescribeSnapshots
+              - fsx:DescribeBackups
+              - fsx:ListTagsForResource
+              - cloudwatch:DescribeAlarms
+              - cloudwatch:GetMetricData
+              - cloudwatch:GetMetricStatistics
+              - cloudwatch:ListMetrics
+              - backup:ListProtectedResources
+              - backup:ListRecoveryPointsByResource
+              - ec2:DescribeSecurityGroups
+              - ec2:DescribeSubnets
+              - ec2:DescribeRouteTables
+            Resource: '*'
+
 Outputs:
   DevOpsAgentRoleArn:
     Description: Role ARN to use with aws devops-agent associate-service.
@@ -555,6 +605,7 @@ Outputs:
         - agentcore-observability-setup: ${EnableAgentCoreObservabilitySetup} (bedrock-agentcore:Get/ListAgentRuntime, xray:GetTraceSegmentDestination, logs:DescribeDeliveries/DeliverySources/DeliveryDestinations/ResourcePolicies, lambda:GetFunctionConfiguration, ecs:DescribeTaskDefinition/DescribeServices/ListTasks, eks:DescribeCluster)
         - agentcore-ops-review: ${EnableAgentCoreOpsReview} (bedrock-agentcore read-only List/Get for runtimes/memories/gateways/browsers/code-interpreters/workload-identities, ec2:DescribeSubnets)
         - sagemaker-ai-ops-review: ${EnableSageMakerAIOpsReview} (savingsplans:DescribeSavingsPlans)
+        - aws-fsxn-operations-review: ${EnableAwsFsxnOperationsReview} (fsx:Describe*/ListTagsForResource, cloudwatch:DescribeAlarms/GetMetricData/GetMetricStatistics/ListMetrics, backup:ListProtectedResources/ListRecoveryPointsByResource, ec2:DescribeSecurityGroups/Subnets/RouteTables)
       Skills covered by AIDevOpsAgentAccessPolicy (no extra policy needed):
         - aws-eks-operations-review, eks-upgrade-readiness, enrich-with-aws-security-agent, crm-production-investigation-guidelines
       No IAM required:

diff --git a/skills/aws-fsxn-operations-review/.skilleval.yaml b/skills/aws-fsxn-operations-review/.skilleval.yaml
@@ -0,0 +1,3 @@
+audit:
+  ignore:
+    - STR-016    # README alongside SKILL.md is intentional
diff --git a/skills/aws-fsxn-operations-review/CHANGELOG.md b/skills/aws-fsxn-operations-review/CHANGELOG.md
@@ -0,0 +1,28 @@
+# Changelog
+
+All notable changes to the `aws-fsxn-operations-review` skill are documented here.
+
+## [1.5.0] - 2026-10-02
+
+Initial public release of the Amazon FSx for NetApp ONTAP Operational Review skill.
+
+- **Scope:** strict read-only, fully automated operational review of FSx for NetApp
+  ONTAP (file system / SVM / volume / whole footprint) across five pillars —
+  Backup, Observability, Operations, Performance, and Security — for production or
+  pre-production. Self-discovers all ONTAP file systems in the account/region.
+- **46 checks** driven entirely by public AWS APIs via `use_aws`
+  (`fsx`, `cloudwatch`, `backup`, `ec2`): `Describe*` / `List*` / `Get*` and
+  CloudWatch metric reads only. No manual steps and no ONTAP CLI.
+- **Evidence guardrails:** every result is based solely on values returned during the
+  run; missing/failed/empty data is reported as "Not evaluated" with a reason rather
+  than a guessed Pass/Fail. No assumptions or inference of unobserved state.
+- **References split by pillar** (`overview.md` + one file per pillar) for reliable
+  context loading, plus a full report template under `assets/`.
+- **Gen2 handling:** computes SSD utilization from per-tier `StorageUsed{SSD}` /
+  `StorageCapacity{SSD}` because Gen2 file systems do not emit
+  `StorageCapacityUtilization`.
+- **SVM root volume** (`JunctionPath = "/"`) is detected from returned fields,
+  excluded only where the metric is invalid (OPS-10), tagged elsewhere, and never
+  treated as a deletion candidate.
+- Least-privilege IAM policy and evaluation results (structure, best-practices,
+  functional) included.
diff --git a/skills/aws-fsxn-operations-review/README.md b/skills/aws-fsxn-operations-review/README.md
@@ -0,0 +1,185 @@
+# Amazon FSx for NetApp ONTAP Operational Review — AWS DevOps Agent Skill
+
+Performs a strictly **read-only, fully automated** operational review of Amazon FSx for NetApp ONTAP
+(FSxN) file systems, SVMs, and volumes — across **5 pillars and 46 checks** — producing a
+severity-ranked **Amazon FSx for NetApp ONTAP Operational Review** report with one recommendation per
+Critical, High, or Medium finding.
+
+Every check runs against **public AWS APIs** (`fsx`, `cloudwatch`, `backup`, `ec2`) via the `use_aws`
+tool. There are **no manual steps, no customer questionnaire, and no ONTAP CLI commands** anywhere in
+the review. ONTAP-CLI-level diagnosis is out of scope for this skill.
+
+## Purpose
+
+Teams running FSxN at scale accumulate posture drift that no single console page surfaces: volumes left
+without a snapshot policy, file systems with no automatic backups, SSD tiers quietly climbing past 90%
+(where tiering promotion stops), security groups open to the world, alarms sitting in ALARM state. This
+skill gives the agent the check definitions, severity model, and report format to assess all of it in
+one pass from native AWS control-plane and CloudWatch APIs, and to return findings ordered by how much
+they matter.
+
+## Operational Posture Review (Production or Pre-production)
+
+This is an **operational posture review** that can be run against any FSxN file system — whether in
+**production or pre-production**. Because every finding is severity-ranked and carries exactly one
+concrete remediation, the report is a prioritized action list: clear the Critical, High, and Medium
+findings to improve posture.
+
+| When | Why |
+|------|-----|
+| Recurring (weekly / monthly) | Catch posture drift on a running file system; successive runs are directly comparable |
+| Ad hoc | Audit a production or pre-production FSxN file system and get a prioritized remediation list |
+| Health check | Full best-practices sweep when something feels off |
+
+## Key Capabilities
+
+- **46 checks across 5 pillars** — Backup, Observability, Operations, Performance, and Security.
+  `references/overview.md` (severity model, status values, Evidence Guardrails, APIs/metrics) plus the
+  per-pillar `references/<pillar>-checks.md` files are the authoritative definition of each check's
+  public APIs, logic, thresholds, severity, and output fields.
+- **Fully automated, public-API only** — every check is answerable from `fsx` / `cloudwatch` / `backup`
+  / `ec2` `Describe*` / `List*` / `Get*` calls. No ONTAP CLI, no questionnaire, no customer interview.
+- **Uniform severity ranking** — every finding is Critical, High, Medium, Low, or Informational, and the
+  Executive Summary ranks them most-severe first with per-pillar pass/warning/fail counts.
+- **One recommendation per Critical/High/Medium finding** — concrete and FSxN-specific, never generic.
+- **Per-check failure isolation** — a denied or failing API becomes an error row on that check and the
+  review continues; an `AccessDenied` is reported as "not evaluated — permission not granted" rather than
+  a false "none found".
+
+## Pillars
+
+| Pillar | Checks | Focus |
+|--------|--------|-------|
+| Backup | 7 | AWS Backup coverage, recovery points, SnapMirror (DP) volumes, FS automatic backups, backup growth, staleness, daily cadence |
+| Observability | 3 | CloudWatch alarm coverage and state |
+| Operations | 11 | Tagging (FS + volume/SVM), unused volumes, maintenance windows, storage efficiency, snapshot policies, deployment type, AD, per-volume capacity, aged snapshots |
+| Performance | 17 | CPU, IOPS, throughput, SSD capacity, tiering, FlexGroup/FlexClone, capacity-pool reads, HA pairs, generation, cache hit ratio, latency |
+| Security | 8 | Security-group ingress/egress and exposure, volume security style, AD integration, encryption/KMS, Multi-AZ routing |
+
+## Prerequisites
+
+### 1. An AWS DevOps Agent Space with the target AWS account
+
+An existing Agent Space with each account you want to review configured as a cloud source, and the
+`use_aws` tool available to the agent.
+
+### 2. IAM permissions
+
+The skill needs read-only access to `fsx`, `cloudwatch`, `backup`, and `ec2`. These permissions are
+delivered repo-wide through [`cloudformation/devops-agent-skill-policies.yaml`](../../cloudformation/devops-agent-skill-policies.yaml),
+which attaches the inline policy `DevOpsAgentSkill-AwsFsxnOperationsReview` to your DevOps Agent role. The
+skill is enabled by default (`EnableAwsFsxnOperationsReview=true`). Deploy or update the stack against your
+agent role:
+
+```bash
+aws cloudformation deploy \
+  --template-file cloudformation/devops-agent-skill-policies.yaml \
+  --stack-name devops-agent-skill-policies \
+  --parameter-overrides ExistingRoleName=<YOUR-DEVOPS-AGENT-ROLE-NAME> \
+  --capabilities CAPABILITY_NAMED_IAM
+```
+
+The skill operates strictly read-only: no `Create*`, `Update*`, or `Delete*` calls, no mounting, and no
+data-plane access of any kind.
+
+### 3. FSxN file systems with activity (recommended)
+
+Performance and capacity checks read `AWS/FSx` CloudWatch metrics, which publish while a file system is
+in use. Reviewing an idle file system produces "No data" rows rather than false findings.
+
+## Limitations
+
+- **Control-plane and CloudWatch metrics only.** The review reports configuration and CloudWatch signals
+  exposed by public AWS APIs. ONTAP-internal state — SnapMirror lag/health, WAFL/aggregate internals,
+  per-volume ONTAP latency counters, FlexCache hit ratios, audit-log and snapshot auto-delete policy — is
+  **not** assessed by the review.
+- **No ONTAP CLI in the review.** CLI-level diagnosis is out of scope for this skill.
+- **Point-in-time.** Findings reflect state at run time; per-check metric windows are fixed.
+- **Large estates may need scoping.** Many file systems across many regions can exhaust the run budget;
+  scope to fewer file systems or pillars if a run times out.
+
+## Design Notes
+
+This skill is a general-purpose FSxN operational review:
+
+- **Fully automated — no manual steps.** No customer questionnaire and no ONTAP CLI "manual
+  verification" items. Every check is answerable from a public AWS API.
+- **Workload-agnostic.** It assesses general FSxN best practices and does not embed assumptions about
+  any specific workload type.
+- **Public AWS APIs only.** Every check calls a public AWS API via `use_aws` — `fsx`, `cloudwatch`,
+  `backup`, `ec2` — with no dependency on any non-public control plane. The per-pillar
+  `references/<pillar>-checks.md` files list the exact public operation and `AWS/FSx` CloudWatch metric
+  for each check.
+
+## Skill Contents
+
+```
+skills/aws-fsxn-operations-review/
+├── SKILL.md                              # skill instructions — scope, check execution, report format
+├── README.md
+├── references/
+│   ├── overview.md                       # severity model, status values, Evidence Guardrails, APIs/metrics
+│   ├── backup-checks.md                  # Backup pillar (7 checks)
+│   ├── observability-checks.md           # Observability pillar (3 checks)
+│   ├── operations-checks.md              # Operations pillar (11 checks)
+│   ├── performance-checks.md             # Performance pillar (17 checks)
+│   └── security-checks.md                # Security pillar (8 checks)
+├── assets/
+│   └── report-template.md                # full tabular report format (Evaluation agent / on request)
+└── evals/
+    ├── evals.json                        # functional eval definitions (skill-eval tool input)
+    ├── additional-permissions.json       # read-only perms for the functional eval's agent spaces
+    └── README.md                         # how evaluation results are generated and stored
+```
+
+The least-privilege IAM for this skill is delivered repo-wide through the shared template
+[`cloudformation/devops-agent-skill-policies.yaml`](../../cloudformation/devops-agent-skill-policies.yaml),
+which adds the `DevOpsAgentSkill-AwsFsxnOperationsReview` inline policy when
+`EnableAwsFsxnOperationsReview` is `true` (the default).
+
+## How to Use This Skill
+
+Recurring posture review of a running workload (the primary use case):
+
+> Run an operational review of our production FSx for NetApp ONTAP file systems and give me the
+> prioritized findings.
+
+Full review of specific file systems:
+
+> Run an FSx for NetApp ONTAP operational review for account 123456789012 in us-east-1, file systems
+> fs-0123456789abcdef0 and fs-0abcdef1234567890.
+
+Single pillar:
+
+> Review just the Performance pillar of my FSxN file systems — CPU, IOPS, SSD capacity, and tiering.
+
+Targeted question that still routes through the check definitions:
+
+> Which of my FSxN volumes have no snapshot policy or use ALL tiering on active data?
+
+## Uploading to AWS DevOps Agent
+
+Package the skill from inside the skill directory so `SKILL.md` sits at the archive root:
+
+```bash
+cd skills/aws-fsxn-operations-review
+zip -qrD ../../aws-fsxn-operations-review.zip . \
+  -x 'README.md' 'evals/*' '.DS_Store' '*/.DS_Store'
+unzip -l ../../aws-fsxn-operations-review.zip
+```
+
+(`README.md` and `evals/` are repo artifacts, excluded from the uploaded skill.)
+
+Upload the zip via the Agent Space Operator Web App Skills page. Target the **Chat tasks** and
+**Evaluation** agent types for on-demand and scheduled review runs.
+
+## Non-production disclaimer
+
+> ⚠️ This skill is sample code, not intended for production use without additional review and testing.
+> It performs read-only operational analysis and makes no changes to your AWS resources, but you are
+> responsible for reviewing the IAM permissions you grant and for validating the findings and
+> recommendations it produces before acting on them.
+
+## License
+
+MIT-0