AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

azure-upgrade

Assess and upgrade Azure workloads between plans, tiers, or SKUs, or modernize Azure SDK dependencies in source code. WHEN: upgrade Consumption to Flex Consumption, upgrade Azure Functions plan, change hosting plan, function app SKU, migrate App Service to Container Apps, modernize legacy Azure Java SDKs (com.microsoft.azure to com.azure), migrate Azure Cache for Redis (ACR/ACRE) to Azure Managed Redis (AMR).

Original source
microsoft
License
MIT
Pinned revision
3f5b461b2314e01c400564051e92469114c5b68f

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

This skill also requires local files or dependencies. Remote reading alone may not be enough.

Ask your assistant

Use Agentic Memory skills to read microsoft--azure-upgrade at revision 3f5b461b2314e01c400564051e92469114c5b68f, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 49

references/global-rules.md

Global Rules

These rules apply to ALL phases of the azure-upgrade skill.

Destructive Action Policy

⛔ NEVER perform destructive actions without explicit user confirmation via ask_user:

  • Deleting apps, services, or resource groups
  • Stopping or disabling the original app/service
  • Overwriting app settings or configuration in the new app
  • Removing the original hosting plan or service tier
  • Modifying DNS or custom domain bindings

User Confirmation Required

Always use ask_user before:

  • Selecting target Azure subscription
  • Selecting target Azure region/location
  • Creating new Azure resources
  • Stopping or deleting the original app/service
  • Modifying custom domains or network restrictions
  • Any irreversible configuration change

Best Practices

  • Always use mcp_azure_mcp_get_azure_bestpractices tool before generating upgrade commands
  • Prefer managed identity over connection strings — upgrades are a good time to improve security
  • Always target the latest supported runtime version — check Azure docs for the newest GA version
  • Keep the original app/service running until the upgraded one is fully validated
  • Use the same resource group for the new resource to maintain access to existing dependencies
  • Follow Azure naming conventions for all new resources

Identity-First Authentication (Zero API Keys)

Enterprise subscriptions commonly enforce policies that block local auth. Always design for identity-based access from the start.

  • Prefer managed identity connections over connection strings/keys
  • Use DefaultAzureCredential in code — works locally and in Azure
  • When using User Assigned Managed Identity, always pass managedIdentityClientId explicitly
  • See service-specific identity configuration in the scenario reference files

Rollback Policy

  • Always document rollback steps before executing upgrade
  • Keep the original app intact and running until upgrade is validated
  • If upgrade fails, guide the user to restart the original app
  • Never delete the original app automatically — always require ask_user

Artifact Output Policy

All migration artifacts (assessment reports, status files, downloaded packages, transient JSON used to transfer settings) must be written under a single, scoped directory — never loose in the workspace root.

Placeholder

| Placeholder | Resolves to | |---|---| | <UPGRADE_DIR> | <workspace-root>/.azure-upgrade/<scenario-id>/ (override: $env:AZURE_UPGRADE_DIR on PowerShell, $AZURE_UPGRADE_DIR on bash) |

<scenario-id> is defined by each scenario reference:

| Scenario | <scenario-id> | |---|---| | Functions Consumption → Flex (Linux or Windows) | source function-app name (e.g. mathewc-test-isolated) | | Java SDK modernization | uses its own location: .github/java-upgrade/{RUN_ID}/ (historical — TODO: converge to <UPGRADE_DIR> in a future pass) | | Redis (ACR/ACRE) → AMR | n/a — routes to external skills, this policy does not apply |

Rules

  1. Resolve <UPGRADE_DIR> once at session start. Honor $AZURE_UPGRADE_DIR if set; otherwise use .azure-upgrade/<scenario-id>/ under the current workspace root. Create the directory if missing.
  2. Announce the chosen directory to the user before the first write, e.g. "I'll save migration artifacts under .azure-upgrade/<scenario-id>/. Say so if you'd prefer a different location." — non-blocking, same pattern as naming/storage choices in the scenario references.
  3. Add .azure-upgrade/ to .gitignore if the workspace is a git repo and the entry is missing. Append once; never modify other entries.
  4. Pass explicit paths to all download/write commands (--file <UPGRADE_DIR>/..., --dest <UPGRADE_DIR>, > <UPGRADE_DIR>/...). Do not rely on the current working directory — terminal sessions reset cwd between calls.
  5. Clean up transient artifacts after the operation that produced them succeeds. Generated files and downloaded packages may contain user code or interim state — delete them once they have been applied. If the operation fails, leave retryable non-secret artifacts in place so the user does not need to recreate or re-fetch them.
  6. Never persist raw secret values to disk — anywhere, including <UPGRADE_DIR>. Connection strings, keys, passwords, and secret-bearing app settings must stay in scoped process memory only. Suppress command output and clear variables immediately after use. <UPGRADE_DIR> is for migration artifacts users already handle as part of normal app development — not raw secrets.