labs/deployment-compliance/skills/deployment-compliance-check/SKILL.md
Checks whether Azure Container App deployments comply with the organization's CI/CD-only deployment policy. Uses three signals: Activity Log caller identity, Docker image labels (tamper-proof), and resource tags. QueryLogAnalyticsByWorkspaceId
npx skillsauth add microsoft/sre-agent deployment-compliance-checkInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Organization Policy All Container App deployments MUST go through the approved CI/CD pipeline (GitHub Actions).
Deployments via Azure Portal, interactive Azure CLI, or PowerShell are non-compliant. Only service principal / managed identity deployments from the CI/CD pipeline are compliant. Non-compliant deployments should be flagged, reported, and reverted (with user approval). This policy ensures every production change is traceable to a code commit, reviewed via PR, and auditable through the pipeline.
How the Pipeline Works GitHub Actions builds the Docker image with immutable compliance labels, pushes to ACR, which fires an Event Grid event. An Automation Runbook (running under a managed identity) picks up the event and updates the Container App via ARM. The key point: GitHub never authenticates to Azure AD directly — all Azure-side auth happens through managed identities inside Azure.
Data Sources Activity Logs in Log Analytics Activity Logs flow to the Log Analytics workspace via diagnostic settings. Use QueryLogAnalyticsByWorkspaceId to run KQL against the AzureActivity table.
To discover the workspace ID if needed:
az monitor log-analytics workspace show --resource-group rg-compliancedemo --workspace-name law-compliance-compliancedemo --query customerId -o tsv Container App Resource Tags Use RunAzCliReadCommands to check tags on the Container App.
Docker Image Labels in ACR The CI/CD pipeline bakes labels into every image at build time (deployed-by, commit-sha, pipeline-run-id, branch, repository, workflow). These are immutable once pushed — they cannot be added or changed after the fact. An image pushed manually (via Portal or docker push) will NOT have these labels.
How to Detect Compliance See compliance_detection.md for the detailed decision tree and well-known app IDs.
Step 1: Query Activity Logs Query the AzureActivity table for Container App write operations. Extract claims.appid and Caller to identify who made the deployment. See compliance_detection.md for the KQL template.
Step 2: Classify each deployment by caller Well-known Azure Portal / CLI / PowerShell app IDs → NON-COMPLIANT Caller contains @ (user principal) → NON-COMPLIANT Known pipeline managed identity → proceed to Step 3 Unknown service principal → INVESTIGATE Caller identity ALWAYS takes precedence over tags.
Step 3: Verify Docker image labels (the tamper-proof check) This is the most important step. Even if the caller is the pipeline's managed identity, the image itself might have been pushed to ACR manually (bypassing the CI/CD build). When that happens, Event Grid still fires, the Automation Runbook still deploys it, and the Activity Log looks legitimate — but the image was never built by GitHub Actions.
To catch this:
Get the currently running image tag from the Container App Retrieve the image config from ACR and check for the expected labels (deployed-by=pipeline, commit-sha, pipeline-run-id, etc.) If labels are missing or invalid → NON-COMPLIANT regardless of caller This closes the "portal push via Event Grid" bypass.
Step 4: Verify resource tags (secondary) Compliant pipelines stamp tags like deployed-by=pipeline, pipeline-run-id, commit-sha, repository. Missing deployed-by tag is additional non-compliance evidence — but tags alone are weak because the Automation Runbook stamps them on every deploy, including ones triggered by manual ACR pushes.
Step 5: Generate compliance report Report should include scan timestamp, time range, total/compliant/non-compliant counts, image label check results, and details of any violations.
Revert Procedures IMPORTANT: Always get user approval before any revert action.
Option A — Reactivate previous Container App revision: list revisions, activate the last known-good one, shift traffic, deactivate the non-compliant revision.
Option B — Re-run the CI/CD pipeline to redeploy the last known compliant image from the approved pipeline.
Notes Activity Logs may take 5-15 minutes to appear in Log Analytics claims.appid values for Portal/CLI/PowerShell are well-known Microsoft constants (see compliance_detection.md) Caller identity is authoritative; tags can be stale from previous deploys Docker image labels are the strongest signal — immutable once pushed to ACR The "portal push" attack path: manual ACR push → Event Grid → Automation → looks compliant but image labels are missing Never revert without user approval
development
Use to package a completed Zava Learning incident analysis for the audience. First present a branded in-thread executive summary (markdown with the before/after visuals inline), then produce the downloadable deliverables — a PowerPoint deck, an HTML email, and a Teams notification — using the Zava corporate template. The calling agent may narrow this to a subset (e.g. only the HTML report). Produces content and artifacts; it does not send them. Assembles the output of rca-analysis, evidence-before-after, recommendations-next-steps, and pr-delivery.
development
Use to package a completed Zava Learning weekly governance audit (NSG / network security, RBAC / least-privilege, or cloud cost) into a single branded, downloadable PowerPoint deck in the Zava house style. The calling audit agent passes its findings (a list of rows with severity) and a short posture summary; this skill renders the deck, applies redaction, and returns the attachment download link. Produces the artifact; it does not send it.
development
Use whenever a Zava Learning investigation produces a durable fix that needs change management — after a GitHub PR is opened for an Infrastructure-as-Code or application code root cause, raise a ServiceNow Change Request referencing the PR and attach the RCA report. The single owner of ServiceNow Change Request and attachment operations.
development
Use whenever you are about to emit operator-visible content — a chat/thread message, a PagerDuty or ServiceNow note, a commit message or pull-request body, or any report artifact (HTML, PowerPoint, Teams card). Deterministically masks secrets, credentials, tokens, private keys, URI-embedded passwords, and PII so they never appear in the thread or in any deliverable. This is a cross-cutting guardrail invoked by the output-producing skills, not a runbook step.